Parallel profiling of genome mutations and multiple chromatin modalities in single cells by scNanoCT&ATAC-seq

Parallel profiling of genome mutations and multiple chromatin modalities in single cells by scNanoCT&ATAC-seq | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Parallel profiling of genome mutations and multiple chromatin modalities in single cells by scNanoCT&ATAC-seq Xiaoying Fan, Dan Su, Zhuobin Lin, Lin Luo, Enze Deng, Qingpei Huang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8125184/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Profiling multiple histone modifications alongside chromatin accessibility at single-cell resolution has been enabled by technologies such as multi-CUT&Tag, scGET-seq, and nano-CUT&Tag. Here, we developed Nanopore sequencing based single-cell CUT&Tag and ATAC co-profiling (scNanoCT&ATAC-seq), which allows for the simultaneous mapping of genome mutations, histone modifications and chromatin accessibility within the same individual cells. scNanoCT&ATAC-seq achieves comparable sensitivity and fragment yield per cell as previous single-modal methods. Anchoring by the ATAC module, a comprehensive epimap that includes multiple histone marks is characterized, generating a cell type specific annotation of chromatin status for genomic regulatory elements, which specified different H3K9me3 modes and identified enhancer-promoter interactions at single-molecular level, specifying genomic structure variations lead to novel regulatory elements in cancer cells. Applying to lung adenocarcinoma, scNanoCT&ATAC-seq clearly captured the tumor subclones and revealed the coordination of genomic and epigenetic regulations. Together, scNanoCT&ATAC-seq reveals the interplay between genomic and multiple epigenetic modalities underlying any cellular processes. Biological sciences/Biotechnology/Sequencing Biological sciences/Genetics/Epigenomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction The genome and epigenome, encompassing transcription factor (TF) binding, histone modifications and chromatin remodeling, are orchestrated to govern cell fate transitions 1 – 3 . Together, these factors determine the regulatory logic underlying cell state changes during development and disease progression. Alterations in genomic structures and chromatin states often precede transcriptional changes 4 , 5 . Foundational principles of chromatin regulatory include the cooperative binding of epigenetic factors, the co-occurrence of synergistic or antagonistic histone modifications, and the dynamic shifts in chromatin accessibility. For example, cis-regulatory elements are typically marked by the presence of multiple active histone marks and high chromatin accessibility, whereas bivalent loci are characterized by the presence of both active and repressive histone marks alongside constrained accessibility 6 – 8 . Such epistatic interactions across epigenetic modalities are crucial for the rapid activation or repression of gene expression. Therefore, comprehensive multimodal chromatin mapping holds promise for predicting cell lineage commitment or cellular state transition prior to overt changes in transcriptional programs. Recent advances in single-cell epigenomic technologies, such as single-cell assay for transposase-accessible chromatin with sequencing (scATAC-seq) 9 , 10 , single-cell CUT&Tag (scCUT&Tag) 11 – 15 , and their multi-modal derivatives have greatly enhanced the resolution of chromatin profiling. Multi-CUT&Tag, for example, enables simultaneously profiling of two chromatin proteins within the same cells by pre-coupling antibodies to pA-Tn5 loaded with distinct barcoded adapters 16 , 17 . Further increasing multiplexing capability, scNano-CUT&Tag employs nanobody-Tn5 fusions to profile up to three epigenetic modalities at single-cell resolution 18 , 19 . However, its reliance on customized nano-antibodies limits broad adoption. Although these methods effectively map euchromatic marks such as H3K27ac and H3K27me3, their performance in capturing heterochromatic regions marked by H3K9me3 remains unclear. Among published single-cell methods, scGET-seq 20 has been shown to probe both heterochromatin and euchromatin by utilizing an engineered transposase fused to the chromodomain of the heterochrmatin protein-1a (HP-1a), offering a more detailed description of genomic copy number variations (CNVs) compared to scATAC-seq. Still, none of these methods can capture structure variations (SVs) and epigenetic modalities in the same individual cells. The combination of long-read sequencing with single-cell assays has opened new avenues for studying genetic regulation. For instance, scNanoATAC-seq 21 combines scATAC-seq with Nanopore sequencing to simultaneously assess chromatin accessibility and genomic SVs. More recently, the same team developed scNanoSeq-CUT&Tag 22 , which enables histone mark profiling in complex genomic regions that are challenging to study with short reads. Nevertheless, the potential of long-read sequencing for simultaneous profiling of histone modifications and chromatin accessibility has yet to be fully explored. In this study, we present scNanoCT&ATAC-seq, a single-cell assay that leverages Nanopore sequencing to simultaneously profile genome variations, histone modifications, and chromatin accessibility within individual cells. Our approach demonstrates comparable sensitivity and accuracy with established single-modality methods, including scNanoATAC-seq and scNanoSeq-CUT&Tag. We show, at single-molecule resolution, that genomic SVs can induce novel enhancer–promoter (E-P) interactions in cancer cells. Applied to a lung adenocarcinoma (LUAD) sample, scNanoCT&ATAC-seq clearly separates tumor and normal epithelial cells based on both genomic and epigenomic features, and further resolves intratumoral subclones. We anticipate that scNanoCT&ATAC-seq will represent a valuable addition to the toolkit for multimodal single-cell profiling technologies. Results scNanoCT&ATAC-seq efficiently maps chromatin accessibility and histone modification landmarks in the same individual cells We coupled protein A/G-fused Tn5 (pA/G-Tn5) and Tn5 with different adaptor sequences, referred to as I5-pA/G-Tn5 and I7-Tn5, respectively. In the scNanoCT&ATAC-seq approach (Fig. 1a), single-cell suspensions are incubated with primary antibodies targeting specific histone modifications or TFs. Afterward, the cells are treated with a secondary antibody and I5-pA/G-Tn5. Following washing out the free I5-pA/G-Tn5, the cells are incubated with I7-Tn5 while Mg 2+ ions are added to activate the tagmentation reactions. Subsequently, individual tagmentated cells are sorted into 96-well plates containing cell lysis buffer and preamplification primers, with each well containing two primer sequences sharing the same 16-bp inner cell barcode. After protein lysis, the DNA fragments undergo a first round of PCR to generate long amplicons. The amplicons with distinct inner cell barcodes are pooled and undergo a second round of PCR to tag them with outer cell barcodes. Ultimately, 600–1000 cell amplicons containing diverse combinational barcodes are pooled together for library construction and sequencing on the Oxford Nanopore platform (Fig. 1a). The median fragment length of scNanoCT&ATAC-seq libraries ranges from 3,528 bp to 4,149 bp (Extended Data Fig. 1a,b). According to the library structure, the histone marks and the open chromatin regions are recognized by I5 ends and I7 ends of the reads, respectively (Fig. 1b). We summarized the percentages of reads with different ends for each cell. In GM12878 cells, approximately 99% of reads were usable, as either end could be recognized by a specific adaptor (Fig. 1c). Moreover, the ratio of I5 to I7 ends varies across combinations with different histone marks. For example, I5-containing reads accounted for only 51.69% in H3K9me3 scNanoCT&ATAC-seq libraries, while they constituted 77.97% in H3K4me3 scNanoCT&ATAC-seq libraries. Reads with heterogenous ends were shorter, indicating crosstalk between open chromatin and histone modifications across the genome (Extended Data Fig. 1c). Compared to long-read sequencing-based single-modality methods, including scNanoATAC-seq 21 and scNanoSeq-CUT&Tag 22 , our multiomics approach achieves a comparable mapping ratio (~ 98%, except for the H3K9me3 group, Extended Data Fig. 1d) and sensitivity (Fig. 1d, up to 41,775 and 39,134 unique reads per cell for ATAC and histone modifications, respectively). With 2,700 to 417,504 reads per cell, the median single-cell genome coverage for scNanoCT&ATAC-seq is 5.84% (Extended Data Fig. 1e). Furthermore, transcription start site (TSS) scores from ATAC modality ranged from 1.02 to 11.70, higher than those from the H3K4me3 modality, while the H3K9me3 modality showed the lowest TSS scores (Extended Data Fig. 1f-g). Specifically, the ATAC signal from scNanoCT&ATAC-seq displayed nucleosome occupancy pattern around TSS, similar to that observed in scNanoATAC-seq (Extended Data Fig. 1h). The fraction of reads in peaks (FRIP) for each modality from scNanoCT&ATAC-seq data was also comparable to corresponding single-modality methods (Fig. 1e). These results suggest scNanoCT&ATAC-seq is highly sensitively in detecting both histone marks and open chromatin regions, with low signal-to-noise ratios, similar to previous single-omics methods. Next, we evaluated the accuracy and reproducibility of scNanoCT&ATAC-seq in profiling multiple chromatin modalities. Libraries from different batches exhibited high consistency within each modality (Extended Data Fig. 2a). The ATAC module demonstrated remarkable similarity across different combinations of histone marks, while H3K4me3, H3K27ac, H3K27me3 and H3K9me3 exhibited distinct patterns (Fig. 1f-g, Extended Data Fig. 2b-c). We merged cells from all replicates (a total of ~ 400 cells per histone mark, Extended Data Table S1 ) and extracted peaks for each chromatin modality, which largely overlapped with the corresponding bulk profiles from the Encyclopedia of DNA Elements (ENCODE) 23 (Extended Data Fig. 2c-d, see methods). Each chromatin modality demonstrated significant positive enrichment in its corresponding bulk annotated region, with H3K9me3 largely excluded from active regions (Extended Data Fig. 3a). Genome browser tracks of merged scNanoCT&ATAC-seq cells indicated that each epigenetic modality closely resembled the ENCODE references for GM12878, while distinct modalities within the same population were markedly different, particularly for active and repressive marks (Fig. 1h). The precision of peak calling was globally higher for active histone marks, remaining stable (over 90%) when the sample size exceeded 50 cells (Fig. 1i). Additionally, around 50% of H3K4me3 and H3K27me3 bulk peaks were recaptured in our datasets (Fig. 1i). Further profile annotation revealed that over 50% H3K4me3 peaks corresponded to promoters, whereas H3K9me3-modified regions were predominantly located in distal intergenic regions (Extended Data Fig. 3b). Similarly, over 60% of unique ATAC, H3K4me3 and H3K27ac peaks identified from scNanoCTA-seq data were enriched at ENCODE candidate cis-regulatory elements (cCREs), while only 20% of H3K9me3 peaks were recognized as cCREs (Extended Data Fig. 3c), indicating efficient and accurate capture of peaks associated with active regulatory elements. Collectively, our analysis demonstrates that scNanoCT&ATAC-seq can efficiently map chromatin accessibility and histone mark occupancy in single cells. ScNanoCT&ATAC-seq accurately captures genomic and epigenetic differences between different cell types Next, we examined the ability of scNanoCT&ATAC-seq to distinguish between different cell types. Unsupervised clustering further revealed that the GM12878 and K562 cells were clearly separated in the Uniform manifold approximation and projection (UMAP) analysis for each modality (Fig. 2a, Extended Data Fig. 4a-b). The two types of cells exhibited a low abundance of common peaks for each epigenetic modality (Extended Data Fig. 4c). The differentially accessible regions (DARs) identified in GM12878 and K562 cells were enriched in pathways related to the B cells and leukocytes, respectively, underscoring the accuracy of scNanoCT&ATAC-seq in capturing the epigenetic characters of individual cell types (Extended Data Fig. 4d). B cell markers such as MS4A1, CD79A and HLA-DRB1 displayed higher intensities of ATAC, H3K4me3 and H3K27ac signals in GM12878 cells, whereas K562 cells showed elevated peak intensities in leukocyte-related genes including BCR, ABL1, GATA1, FAM30A (Fig. 2b-d, Extended Data Fig. 4e). Interestingly, these markers also exhibited scattered signals for repressive marks H3K27me3 and H3K9me3 in their corresponding cell types, suggesting that highly active regions are also dynamically modified by repressive histone marks (Fig. 2b-d). Since scNanoCT&ATAC-seq also covers normal genome fragments by the long reads (Fig. 1b), we evaluated its capacity to detect SVs–including insertions, deletions, inversions and duplications–in K562 cells. Compared to ONT bulk sequencing data 21 , scNanoCT&ATAC-seq data effectively restored the identified control SVs, within highly consistent breakpoints (Extended Data Fig. 5a-b and Extended Data Table S2). The precision of SV detection increased to over 75% when supported by just three cells (Fig. 2e). Using fewer than 400 cells (Extended Data Table S1 ), nearly 90% of the control SVs were recalled, with 3–4 supporting cells being the recommended criteria to achieve both high precision and recovery for SVs (Fig. 2e). Notably, the classical BCR-ABL1 24 fusion ranked in the top two among all identified K562 variants (Fig. 2f, Extended Data Table S2). We also observed that the most frequent SV occurred in intergenic regions (chr13: 108,009,126 - chr9:FAM78A), which was also detected in the control and being validated unique in K562 cells (Fig. 2f, Extended Data Fig. 5c-d). With superior genome coverage over short-read sequencing based singe-cell epigenome sequencing methods, scNanoCT&ATAC-seq revealed a substantial abundance of CNVs unique to K562 cells, distinct from those in GM12878 cells (Fig. 2g, Extended Data Fig. 6). These collective analyses demonstrate that scNanoCT&ATAC-seq efficiently captures specific genomic and epigenomic characteristics in single cells. Functional genomic characterization with combinatorial chromatin modalities in the same cells As scNanoCT&ATAC-seq generates ATAC signal for every cell together with one of the four histone marks in the current dataset, we directly integrated all the single-cell profiles using ATAC modality (Fig. 2a). Then we assigned the chromatin states by ChromHMM 25 at pseudobulk level. A total of 10 chromatin states were obtained by running the Baum-Welch step of ChromHMM (Fig. 3a, Extended Data Fig. 7a-c), which grouped into: E1, flanking TSS (strongly enriched for H3K4me3 and weekly for H3K27ac/ATAC); E2, active TSS (strongly enriched for H3K4me3/H3K27ac/ATAC); E3, strong transcription (enriched in none of the epi-modalities but around TSS); E4, genic enhancer (enriched for H3K27ac/ATAC); E5, unmarked TSS (strongly enriched for only ATAC); E6, bivalent TSS (enriched for H3K27me3/ATAC); E7, weak repressed polycomb (weekly enriched for H3K27me3); E8, Repressed polycomb (enriched for H3K27me3 and low ATAC); E9, quiescent states (with no epigenetic signals detected); E10, heterochromatic states (enriched for H3K9me3). The distribution of most genomic elements slightly changed between GM12878 and K562 cells, especially for the E4 genic enhancers, which showed elevated participation in genic region in K562 cells (Extended Data Fig. 7a-c). In consistent to the element definition, E1 to E5 contained higher proportions of expressing genes and the general expression levels were higher. On the contrary, E6 to E10 genes showed hardly expressions (Fig. 3b, Extended Data Fig. 7d). Thus we categorized E1 to E5 as active chromatin regions, and E6 to E10 as repressive chromatin regions. The active histone marks, H3K4me3 and H3K27ac, showed much higher signals than the repressive histone mark H3K27me3 in the active regions. While H3K27me3 signal was much higher than H3K4me3 and H3K27ac signals in the repressive regions (Fig. 3c, Extended Data Fig. 7e). These results indicate that the comprehensive use of the five epigenetic modalities is accurate in defining the chromatin states. Previous study developed scChromHMM 15 , 26 , which assigns the chromatin states at single-cell resolution with bioinformatically interpolated multiple histone marks. We tested the performance of scChromHMM in our scNanoCT&ATAC-seq data. The analysis returned the active and repressive probability for each 200 bp window across the whole genome in every single cell (see methods). The averaged active scores were postively correlated with ATAC, H3K4me3 and H3K27ac signals, while the averaged repressive scores were positively correlated with H3K27me3 signal (Fig. 3d). As large amount of genomic regions contains both active and repressive marks, we used active to repressive (A/R) score to measure the final transcriptional activity of each element. The bulk level defined active regions, E1 to E5, all showed larger A/R scores, while E6 to E10 had low A/R scores (Fig. 3e). We further divided genes to four groups based on their expression levels within each cell type, and compared the A/R scores of each group of genes. The A/R scores were significantly elevated with increased gene expressions. These data suggests that scNanoCT&ATAC-seq allows accurately measuring the chromatin transcription activity at single cell level. We further evaluated whether the A/R score could capture cell type specific transcriptional regulations by unsupervised clustering of all the single cells. The GM12878 and K562 cells could be clearly separated by A/R score, without batch effects from different modalities (Fig. 3g, Extended Data Fig. 7f). The classical markers of the two cell types, such as MS4A1, HLA-DRB1 for GM12878 and BCR, HBG2 for K562, showed higher A/R scores in single cells of the corresponding populations (Fig. 3h, Extended Data Fig. 7g). IGV track of the cell-type specific genes revealed higher intensity of both active and repressive signals in the corresponding cells, but the former was higher (Fig. 3i). Two modes of H3K9me3 were identified. H3K9me3 is regarded as a landmark of repeat-rich constitutive heterochromatin 27 . Interestingly, according to the chromatin states annotation, H3K9me3 shows high priority in both E2 and E10 states (Fig. 3a, Extended Data Fig. 8a), which are annotated as active TSS and heterochromatin, respectively. These two categories of peaks represents two different modes of H3K9me3 modification (Fig. 4a). In both GM12878 and K562 cells, the E2-H3K9me3 peaks are short, highly accessible, gene rich and transcriptional active, while E10-H3K9me3 peaks are typical heterochromatin representative (Fig. 4b-d, Extended Data Fig. 8b-d). These ATAC peak overlapped H3K9me3 peaks were also observed in the bulk ChIP-seq data, indicating they are real regulations instead of artificial from scNanoCT&ATAC-seq (Extended Data Fig. 8e). In addition, the E2-H3K9me3 peaks mainly distributed close to the TSS, showing more cell-type specific characters (Extended Data Fig. 8f-g). Among these E2-H3K9me3 peaks, over 90% are overlapped with the cCREs and over 70% are intersected with active genome compartment (Extended Data Fig. 8h-i). Single-molecular analysis revealed highly interaction between the E2-H3K9me3 peaks and cis-regulatory elements (Extended Data Fig. 8j), suggesting this mode of H3K9me3 highly related to active gene regulation. To further identify the active regulatory characters of the E2-H3K9me3 modification, we compared the enriched TFs in GM12878 and K562 differently detected peaks (Extended Data Fig. 9a). In both cell types, the enriched motifs are highly consistent between ATAC and E2-H3K9me3 modules, resembling cell-type related TFs (Fig. 4e-f). For example, IRF1 as a master regulator in GM12878, showing top activity by ATAC module analysis, also ranked in the top list of E2-H3K9me3 enriched motifs. IRF1 also showed significant higher gene score, ATAC motif score and H3K9me3 motif score in GM12878 cells (Fig. 4e, Extended Data Fig. 9b-d). In contrast, GATA1 is detected as top regulator by both ATAC and E2-H3K9me3 in K562 cells. On the other hand, the classical E10-H3K9me3 modified regions are quite different to the regulatory elements, showing lower cell-type specificity (Extended Data Fig. 9b-d). All these results indicate that H3K9me3 not only constitutes heterochromatin, but also participates in active gene regulation. Accurate profiling of H3K9me3 modifications in transposon elements (TEs) TEs are evolutionary important elements, which accounts over half of the human genome 28 . They have been reported to play as regulatory elements through epigenetic modifications, especially H3K9me3, in both developmental and disease contexts 29 – 31 . However, previous NGS-based epigenetic profiling could not accurately assign the histone modification status of each TE copy on the genome, due to the high similarity of the sequences. Recently published scNanoSeq-CUT&Tag 22 has claimed better coverage on full-length L1Hs profit from long read length of TGS. For all TE classes, there was no significant difference in the number of TE copies covered by classical NGS-based bulk ChIP-seq, scNanoSeq-CUT&Tag, and ScNanoCT&ATAC-seq (Fig. 4g, Extended Data Fig. 9e). However, for relative longer TE copies, the NGS-based bulk ChIP-seq detected lower H3K9me3 signals, indicating incomplete coverage. Specifically, for > 200bp SINEs, > 2000bp LINEs and > 1000bp LTRs, the two TGS based methods showed clearly higher H3K9me3 Scores (Fig. 4h). For example, we observed no read coverage on the long LINE copies within MS4A5 and downstream MS4A1 in the ENCODE ChIP-seq data. Thus, the TGS-based profiling exhibits obvious advantages in exploring the epigenetic status of these repetitive elements in the genome, even at single-cell resolution. Then we checked the distribution of E2- and E10-H3K9me3 in the TE elements. As expected, in both GM12878 and K562 cells, the distribution ratios of E10-H3K9me3 peaks across the TE superfamilies were roughly the same, and the LINE accounted for the largest proportion 32 (Extended Data Fig. 9f). In comparison, the distribution of E2-H3K9me3 on different TE superfamilies switched dramatically from GM12878 to K562 cells. Further enrichment analysis also supported similar distribution of E10-H3K9me3 in different TE families, significantly enriched in L1, ERV1, ERVK and SVA (Fig. 4i). For the E2-H3K9me3, they highly enriched in MIR in GM12878 cells, while in ERV1, ERVK and SVA in K562 cells. This suggests that the regulation of active H3K9me3 modifications may also contribute to tumor development. Evaluating E-P interaction at single-molecular level In similar to the previous TGS-based single-cell epigenome sequencing methods 21 , 22 , scNanoCT&ATAC-seq also identifies the co-presence of two regulatory elements as they were captured on the same reads. As our integrated single-cell multi-epigenetic modalities accurately defined chromatin states in each cell population (Fig. 3a), we further focused on enhancers and promoters (see methods, Extended Data Fig. 10a), the most fundamental regulatory elements for gene transcription. We directly captured the interacting E-P pairs being supported by inter-reads spanning the corresponding peaks, and calculated the interaction score of each interacting enhancer by normalizing the supported reads to the total reads supporting the promoter (Fig. 5a). For both GM12878 and K562, we detected over ten thousand promoters and over thirty thousand enhancers, with the promoters largely shared (45.3%) compared to the enhancers (17.8%), indicating the enhancers showing higher degree of cell type specificity (Fig. 5b). GO analysis of the cell type specific promoters revealed lymphocyte differentiation pathways for GM12878 and myeloid cell differentiation pathways for K562, highly match the identities of the two types of cells (Fig. 5c). On the contrary, the common promoters enriched in basic pathways such as RNA metabolic and histone modification (Extended Data Fig. 10b), suggesting high accuracy of our element definition. In both cell types, the E-P pairs showed distance less than 10kb, while the none interactive enhancers often displayed longer distance to the nearest promoters (Extended Data Fig. 10c). Therefore the length of the scNanoCT&ATAC-seq library fragments limits the detection of long-distance E-P interactions. Within the limited distance (mostly less than 15kb), 68.29% promoters were detected with interacting enhancers (Extended Data Fig. 10d). The most interactive enhancers ranged around 2.5kb from the promoters (Extended Data Fig. 10e). Both the number of interacting enhancers and the global interacting frequencies regulated the transcriptional activities (Extended Data Fig. 10f). Further comparison of the E-P pairs between different cell types revealed that the vast majority enhancers regulating cell type specific promoters were also cell type specific (Fig. 5d, Extended Data Fig. 10g). Interestingly, even for the common promoters, they largely had different interacting enhancers in different cell types (Fig. 5d). Although cell type specific promoters showed the largest difference in E-P interaction frequency, the common E-P pairs had the highest interaction frequency in both cell types, suggesting that the conserved transcriptional regulation is the strongest (Fig. 5e). Theoretically, cell type-specific enhancers regulating the same promoters can serve as new therapeutic targets for disease treatment, especially for tumor treatment. For example, the lysine-specific demethylase KDM8 was commonly highly expressed in both K562 and GM12878 cells, while the upstream regulatory enhancers were completely different in the two types of cells (Fig. 5f, E20546 in K562 and E20547 in GM12878). Therefore, targeting E20546 should be specific to modulate KDM8 expression in K562 cells, without effect on GM12878 cells. As scNanoCT&ATAC-seq profiles the SVs as well as E-P interactions in the same cells, we figured out how genomic variations contribute to abnormal regulatory interactions in cancer. FAM78A, as a tumor associated gene highly expressed in hematologic disorders in the Human Protein Atlas, fused with chr13 in K562 cells (Extended Data Fig. 5c-d). Our data also captured E-P interactions between the chimera region (Fig. 5g). To prove the new enhancer promote FAM78A expression in cancer cells, we deleted the enhancer region using CRISPR system and measured the expression of FAM78A and its upstream gene on chr9 (Fig. 5h). The results validated the new E-P interaction formed by genomic translocation, demonstrating that scNanoCT&ATAC-seq can effectively capture the coordinated changes in genomic variations and regulatory elements in single cells. Identifying genetic and epigenetic changes of tumor cells in LUAD We applied scNanoCT&ATAC-seq to a LUAD sample to identify how genomic alternations cooperate with chromatin accessibility and H3K9me3 in tumorigenesis (Fig. 6a). A total of 1,920 cells were sequenced, and after quality control, we obtained 1,556 cells with matched ATAC and H3K9me3 data (see methods, Extended Data Table S1 ). Using the ATAC module, we clearly identified 6 groups of cells: one cluster of normal epithelial cell, two clusters of tumor cells, one cluster of endothelial cells, one cluster of fibroblasts and one cluster of immune cells (Fig. 6b-c, Extended Data Fig. 11a-b). The tumor cells revealed the largest difference to the other cell types, while the non-epithelial cells were relatively mixed at H3K9me3 level, suggesting dramatic changes during tumorigenesis. Then we analyzed the CNVs of each single cell, confirming copy number abnormalities in tumor cells (Fig. 6d). Meanwhile, the tumor cells were separated into two subclones based on the CNV pattern, which exactly corresponded to the two tumor clusters based on ATAC module. The tumor_1 cells gained more copies in the chromosome 5q32-q35 (Fig. 6d). The chromosome 8q region is a common variant area in LUAD 33 – 35 . Here, significant copy number increases of q22-q24 were also detected in both subclones of tumor cells. At single-cell level, we observed distinct CNVs from the two subclone of cells in this region (Fig. 6d). Moreover, we found the ATAC signals were highly related to the CNVs. For example,the oncogene MYC, which increased the copy number in both clones, showed higher ATAC signal in both clones (Fig. 6d-e, Extended Data Fig. 11c). CHRAC1, which significantly related to the survival of LUAD patients, gained genomic copies in tumor_1 cells, and only these tumor cells showed increased chromatin accessibility. On the other hand, the LUAD oncogene EIF3H, gaining genomic copies in tumor _2 cells, exhibited increased ATAC signals in these cells (Fig. 6d-e, Extended Data Fig. 11c). These results indicate that the epigenetic heterogeneity of tumor cells is highly consistent with the heterogeneity of genomic variations. Considering that the genomic copy number dosage may affect the calculation of the epigenetic signal value 36 , we discussed the chromatin accessibility changes by comparing to the genomic copy number changes (Fig. 6f). Take Tumor_1 cells as example, 277(7.89%) ATAC peaks showing comparable fold changes (FCs) in ATAC signal than copy number FCs, indicating these epigenetic changes are just dosage effect from genomic variations (Fig. 6g, Extended data Table S3). 2,708(77.13%) and 526(14.98%) peaks had higher and lower FCs in ATAC signal than copy numbers respectively, revealing de novo remodeling of these elements. The genes associated with the active increase of chromatin accessibility in tumor cells were mainly related to ERK1 and ERK2 cascade, Ras protein signal transduction and regulation of GTPase activity, supporting enhanced proliferation and metastasis facilitating tumor development. On the contrary, some RNA regulatory genes such as POLR2K, RNF19A and PABPC1, exhibited a decrease in chromatin accessibility-dependent dosage compensation in the copy number gaining regions. Thus we can clearly distinguish the coordinated regulation of the genome and epigenetic modifications in tumor cells using scNanoCT&ATAC-seq. The two tumor subclones did not show significant differences in H3K9me3 (Fig. 6b), so we merged these cells to analyzed how H3K9me3 changes in tumorigenesis. The H3K9me3 peaks in either tumor or normal cells were defined as E2-H3K9me3 and E10-H3K9me3 according to the accessibility (Extended Fig. 11d-e). For the changes of each type of H3K9me3 modification from normal to tumor cells, we analyzed how they related to the TEs (Extended Data Fig. 11f-g). The remodeling heterochromatin largely associated with LINE1 and ERVK elements, where most copies lost E10-H3K9me3 modification in tumor cells. This observation is consistent with the previous findings that tumorigenesis is accompanied by re-activation of heterochromatin 27 , 37 . Meanwhile, significant amount of LINE1 and LTR elements also gained H3K9me3 modification in tumor cells, implying highly precision of TE regulation (Extended Data Fig. 11g). Together, these results indicate the important role of H3K9me3 in LUAD tumorigenesis, and regulation of this modification is highly concentrated in TE elements such as LNE1 and ERVK. Discussion Histone modifications and chromatin structure provide regulatory diversity that is crucial during differentiation and homeostasis. Leveraging long-read sequencing, scNanoCT&ATAC-seq fundamentally advances single-cell multi-omics by enabling the simultaneous co-profiling of structural genomic variations (CNVs/SVs), multiple histone modifications (including traditionally challenging heterochromatic marks like H3K9me3), and chromatin accessibility within individual cells. This represents an opportunity to not only identify regions that exhibit cell-type-specific accessibility but also highlight elements whose acquisition of activating, repressive or heterochromatic signatures varies within a heterogeneous population. Our method first integrates data from multiple scNanoCT&ATAC-seq experiments together into a common manifold, generating coassay profiles for four histone modifications together with chromatin accessibility within individual cells. This allowed us to systematically define the chromatin states using ChromHMM 25 , and comprehensively assess whether each genomic element is activated or repressed in single cells. One major challenge in human genetics is the prediction of the functional consequences of SVs on gene function and cis-regulatory networks 38 . A key strength of scNanoCT&ATAC-seq lies in the concurrent acquisition of genetic and epigenetic data in single cells. This allows direct correlation of somatic mutations, e.g., SVs and indels, with local chromatin states, providing further insights into chromatin dynamics under specific genomic context. Specifically, we extract genomic mutations and identify the transitions in chromatin states between different cell types, thereby determining in which target cells the regulation of genetic variations occurs for diseases, and what functional elements of the target cells are affected. Similarly, we identified which cis-elements regulate a specific gene in a certain cell population through single-molecule E-P interactions. By comparing E-P interactions in different cells, we found that genes expressed in different cells have enhancer-shift situations, which might be caused by changes in the TF network under different cell states. This method advantage may help in the future development of specific targets for diseases such as tumors, for example, selecting conserved genes for cell survival but unique enhancers used in tumor cells as targets, which may specifically kill tumor cells while reducing side effects. However, the current single-molecule E-P interaction analysis also has certain limitations. Since the fragments recovered from the current library are basically within 10kb, it is difficult to capture E-P interactions over long distances. Fully resolving such architecture may require integration with complementary methods like Hi-C or long-read chromatin conformation capture. Our data also reveal unexpected biological complexity, particularly regarding H3K9me3. We identified two functionally distinct modes: a classical repressive form (E10) in inaccessible, gene-poor heterochromatin, and a novel, active-associated form (E2) co-localizing with H3K4me3, H3K27ac, and accessible chromatin at promoters. These E2-H3K9me3 peaks bind cell-type-specific TFs (e.g., IRF1 in GM12878, GATA1 in K562) and show strong enrichment at cis-regulatory elements, suggesting a potential role in fine-tuning active transcription. Both types of H3K9me3 show dramatic changes in tumorigenesis, and these regulations highly related to TE elements. The mechanistic basis for this paradoxical H3K9me3 function and their role in developmental and disease context demands focused future investigation. Previous studies investigated the epigenetic regulations in cancer cells based on just one modality, such as ATAC-seq, ChIP-seq, etc. Theoretically, the calculation of these epigenetic signals is performed in euploidy. Tumor cells usually acquire large numbers of chromosomal CNVs, and such variations are overlooked in the analysis of epigenomics. By co-profiling of the genome and epigenome in single cells using scNanoCT&ATAC-seq, we show that many tumor-related epigenetic changes result from the dosage effect of genomic copy number. Besides, we propose gene activity maintaining under epigenetic-dependent dosage compensation, where the chromatin becomes less open with the increase in copy number. The epigenetic changes obtained by eliminating the influence of dosage may be more meaningful for therapeutic stratification of patients. Although only one histone mark could be detected together with ATAC in the current version, several improvements can increase the histone modification modalities, i.e., using nanobody-based secondary antibody fused with pA/G-Tn5 18,19 , or pre-conjugating antibodies against histone marks with uniquely barcoded pA/G-Tn5 complexes 39 . Besides, employing engineered reader domain-Tn5 fusions (e.g., HP1a chromodomain for H3K9me3) 20 can also bypass antibody limitations and improve the sensitivity. In summary, scNanoCT&ATAC-seq provides an unprecedented lens to examine the interplay between the genome and multi-layered epigenome in single cells. Its ability to concurrently map mutations, histone modifications, and chromatin accessibility opens new avenues for dissecting cellular heterogeneity in development and disease. Future refinements in scalability, modality diversity, and integration with spatial or 3D genomic data will further cement its role as a transformative tool in precision biology. Methods Ethics statement This study was approved by the Ethics Committee of the First Affiliated Hospital of Guangzhou Medical University (Approval No: ES-2025-034-01). Sample tissues were collected from a LUAD patient with appropriate informed consent. Cell culture and collection K562 and GM12878 cells were cultured in RPMI 1640 medium (Gibco, C11875500CP) supplemented with 10% fetal bovine serum (Procell, 164210-50) 2 mM L-glutamine (Gibco, 25030081) and 1% penicillin/streptomycin (Gibco, 15140122) at 37℃ in a humidified 5% CO₂ atmosphere. Both suspension cell lines can harvest directly by centrifugation at 200g for 5min, while GM12878 cells required additional pipetting to dissociate due to their naturally tendency to form multicellular aggregates. The K562-Cas9-BFP transgenic cell line was obtained from Dr. Nian Liu's laboratory (Tsinghua University). Cells were cultured in the normal K562 medium supplemented with 1% Non-Essential Amino Acids (NEAA; Gibco, 11140-050). Single-cell suspension preparation of the LUAD tissue Surgically resected primary tumor and paired adjacent lung tissues of a LUAD patient were immediately preserved in pre-chilled MACS Tissue Storage Solution (Miltenyi Biotec, DS_130-100-008) and transported to the laboratory on ice. Tissues were minced into small pieces, followed by enzymatic dissociation in 5 ml DMEM/F12 medium (Gibco, #11330500BT) containing 2mg/mL collagenase D (Worthington, #LS004196) and 20U/mL DNase I (Worthington, #LS002004) at 37°C for 30–45 min with gentle shaking to ensure uniform dispersion every 5–10 min. After collagenase digestion, the tissue was thoroughly triturated using 1 ml pipette tip and allowed to settle for 2 min at room temperature. The supernatant containing liberated cells was transferred to a new tube and kept on ice. The remaining undigested tissue clumps were subjected to a second round of enzymatic digestion with 2ml 0.25% trypsin (Invitrogen, 25200056) at 37°C for another 15–30 min. Inactivated trypsin by adding equal volume of 10% FBS plus DMEM/F12 medium and gently pipette to dissociate cells. Passed all the cell suspension through a 40 µm cell strainer. After red blood cell lysis, the cell suspension was subjected to negative selection using CD45 MicroBeads (Miltenyi Biotec, #130-045-801) to enrich non-immune cells according to the manufacturer's protocol. After depletion of immune cells, both primary and adjacent derived cells were used for scNanoCT&ATAC-seq library construction with antibody targeting H3K9me3. Antibodies The following antibodies were used in our experimental procedures: rabbit anti-H3K4me3 (Millipore, 07-473, lot 3746382, 1:50 dilution), rabbit anti-H3K27ac (Abcam, ab177178, lot 1044138-34, 1:100 dilution), rabbit anti-H3K27me3 (Cell Signaling, 9733S, lot 16, 1:40 dilution) and rabbit anti-H3K9me3 (Abcam, ab8898, lot 1063772-1, 1:100 dilution). I5-pA/G_Tn5 and I7_Tn5 Loading The single-stranded I5 adapter (TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG) and the Tn-ME (phos-CTGTCTCTTATACACATCT-NH3) were annealed at equimolar concentrations to form a 50 µM double-stranded I5-adapters. For assembly of the I5-pA/G_Tn5, 20 2 µL pA/G-Tn5 transposase (Vazyme, S604) and 3.5 µL I5-adapters were mixed with 14 µL coupling buffer by gently pipetting, and incubated at 30°C for 1 h. Similarly, the single-stranded I7 adapter (GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG) and the Tn-ME sequence were annealed at equimolar concentrations to form a 50 µM double-stranded I7 adapters. For assembly of the I5 _Tn5, 20 µL Tn5 transposase (Novoprotein, M045) and 4 µL I7-adapters were mixed directly by gently pipetting, and incubated at room temperature for 1 h. ScNanoCT&ATAC-seq experimental workflow 200,000-500,000 viable cells were collected. First, single-cell suspensions were incubated with primary antibodies against specific histone modification at 4°C overnight. Subsequently, these cells were incubated with prebound second antibody tethered I5-pA/G_Tn5 transposase to recognize and amplify the primary antibodies. After washed out excess antibody and I5_pA/G-Tn5 thoroughly, cell pellets were incubated by I7_Tn5 transposase targeting accessible chromatin sites and both I5_pA/G-Tn5 and I7_Tn5 were activated by Mg2 + ions at 37°C for 1 h. Tagmentation was stopped by 20mM EDTA and stained with 10 µg/mL DAPI, the nuclei were immediately sorted using a flow cytometer. Single nucleus was deposited into each well of a 96-well plate pre-loaded with 2 µL of lysis buffer containing 50mM Tris-HCl (pH 8.0), 50 mM NaCl, 1.5 µM of both the I5 and I7 inner barcode primers. Nuclei were lysed to release DNA fragments at 50℃ for 60 min. SDS(contained in the lysis buffer which would inactivate the subsequent PCR polymerase) was quenched with 2 µL 10% Tween20. First round of PCR amplification was performed directly in 10 µL reaction buffer each well containing 0.25U Tks Gflex DNA Polymerase (TAKARA, R060A) as follows: 68℃ for 10min; 95°C for 1 min; 14 cycles of 94°C for 15s, 63°C for 30 s, 68°C for 8 min and finally 68°C for 5 min. PCR products of each row with different inner barcodes were pooled and purified with 0.7X AMPure XP beads (Beckman Coulter; A63882) and eluted with 21 µL H2O. The second round of amplification was carried out in 50 µL reaction mixture with 0.25U Tks Gflex DNA Polymerase and 1 µM different Outer barcode Primers. The PCR program was as follows: 95°C for 1 min, then 5 cycles of 94°C for 15 s, 63°C for 30 s, 68°C for 8 min and finally 68°C for 5 min. After two rounds of PCR amplifications, DNA fragments derived from different cell in each well of 96-well plate were tagged with different inner and outer barcode combination, pooled and purified with 1X AMPure XP beads into a single library. Libraries from 8–10 plates with different barcode combination were pooled together, preceded to ONT library construction with sequencing kit (Oxford Nanopore Technologies, SQK-LSK114) following the manufactures’ instruction and sequenced on one PromethION Flow Cell R10 Version (Oxford Nanopore Technologies, FLO-PRO114M). ScNanoATAC-seq library preparation and sequencing scNanoATAC-seq procedure was adapted from previous work by Hu et al . 21 . Briefly, 200,000 viable cells were lysed with 200 µL Omni-ATAC lysis buffer as described by Corces et al. 40 on ice for 3 min and rinsed by adding 1 ml of ATAC-seq wash buffer. Immediately, nuclei were resuspended with 50 µL tagmentation mixture with 1 µM I7_Tn5 transposase and incubated at 37°C, 800 rpm for 30 min. Transposition was stopped by 20mM EDTA. Nuclei were stained with DAPI and sorted immediately into each well of a 96-well plate pre-loaded with 2 µL lysis buffer with only 1.5 µM I7 inner barcode primers. The subsequent treatment and library construction steps were identical to those described above for the scNanoCT&ATAC-seq. ScNanoSeq-CUT&Tag library preparation and sequencing 200,000-500,000 viable cells as single cell suspension were incubated with primary antibodies against specific histone modification at 4°C overnight. Subsequently, these cells were incubated with prebound second antibody tethered I5-pA/G_Tn5 transposase to recognize and amplify the primary antibodies. After washed out excess antibody and I5_pA/G-Tn5 thoroughly, cell pellets were incubated by 50 µL tagmentation buffer with 10mM Mg2 + to activate transposition at 37°C, 500 rpm for 1 h. After the reaction was terminated by the addition of 20 mM EDTA, the nuclei were stained with DAPI and immediately sorted into a 96-well plate. Each well was pre-loaded with 2 µL of lysis buffer supplemented with only 1.5 µM I5 inner barcode primers. The subsequent treatment and library construction steps were identical to those described above for the scNanoCT&ATAC-seq. Bulk NGS-based CUT&Tag-seq Briefly, 50, 000 cells were collected freshly and centrifuged at 300g for 5 min at room temperature. Cells pellets were incubated with activated ConA beads Pro and proceeded to the following antibody incubation and pA/G-Tnp transposition. Targeted DNA fragments were extracted for library preparation using the Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro (Vazyme, TD904). All NGS libraries were sequenced using SURFSeq 5000 sequencer with PE150 reads. Functional validation of the SV-derived FAM78A E-P interaction After the identification of chromosome fusion event of chr13:108M and chr9: FAM78A, we annotated active enhancers and promoters to fide out that enhancer E15695 may form novel E-P interactions with the promoter of FAM78A to regulatory the expression of this gene. We designed two pairs of single guide RNAs (sg13 and sg24) using the online tool ( http://crispor.org ) targeting the flanking regions of E15695. Oligonucleotides were synthesized by Sangon Biotech (Shanghai, People's Republic of China). The tandem sgRNA cassette was introduced into the lentivirus vector (pLV3-Amp_hU6-esgRNA_EF1a-NeoR-T2A-EGFP-backbone) using multiplex PCR followed by homologous recombination. Lentivirus was produced in 293T cells, and the concentrated virus was then used to infect the K562-Cas9-BFP cells. GFP and BFP double positive cells were sorted by FACS and continued to be cultured for subsequent mRNA extraction and RT-qPCR detection for the quantification of FAM78A gene expression. RAPGEF1 gene located upstream of the chr9: FAM78A breakpoint was also quantified as a non-regulated control. Reverse-transcription and real-time quantitative PCR Cells were collected to extract RNA using RNeasy Mini Kit (Qiagen, 74104). mRNA was reverse-transcribed into cDNA using Maxima H Minus Reverse Transcriptase (ThermoFisher Scientific, EP0753). Real-time quantitative PCR was performed using ChamQ SYBR qPCR Master Mix (Vazyme, Q311-02) according to the manuals on Quantagene q225 real-time PCR system (Kubo Technology) in 10 µL reaction mixture with 3 replicates. Data preprocessing For scNanoCT&ATAC-seq data, Nanopore sequencing raw signals were converted to DNA sequences using the high-accuracy model “dna_r10.4.1_e8.2_400bps_hac” of Dorado ( https://github.com/nanoporetech/dorado/ ) v0.2.0 software. Reads with quality scores less than 7 were discarded. To demultiplex single-cell files from raw sequences, we used the nanoplexer ( https://github.com/hanyue36/nanoplexer ) v0.1 software with default parameters. Demultiplexing was performed twice on the outer and inner cell barcodes. We used the cutadapt 41 v4.1 software with a minimal sequence length of 12 bp and a maximum error rate of 0.2 to trim the adaptors sequentially from both ends of the reads and record the adaptor information. The adaptor sequences include ‘GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG’ employed to tag transposase-accessible chromatin and ‘TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG’ employed to tag chromatin modifications. After trimming the cell barcodes and adaptors, reads with read length less than 100 bp were discarded. Reads were aligned to the human reference genome (GRCh38) using minimap2 42 v2.26 with parameters “-ax map-ont --MD”. Alignments were subjected to sorting by coordinate, filtering by a minimal mapping quality (MAPQ) of 30, and PCR duplicate removal using SAMtools 43 v1.13. For in-house scNanoATAC-seq and scNanoSeq-CUT&Tag data, the procedures and parameters of basecalling, demultiplexing, adaptor trimming, mapping and quality control were identical to those described above. For bulk CUT&TAG-seq data, we used the Trimmomatic 44 v0.39 to trim reads and remove adaptor sequences of Illumina platforms with the parameters: PE MINLEN:20 CROP:100. Read mapping were performed using bowtie2 45 v2.4.2 with parameters “--mm -X2000”. Low-quality alignments and duplicated alignments were also removed using SAMtools. Extraction of chromatin accessibility and modification signals The two ends of the alignments corresponded to open or modified chromatin sites. The aligned bam files were converted into bed format files using the ‘bamtobed’ command of BEDtools 46 v2.30.0. The genome coordinates of two ends were extracted using the ‘flank’ command of BEDtools with a 1 bp flanking size. By integrating the prior cell barcode and adapter information, we derived two chromatin mark signals for each cell, which were output to fragment files in bed format. For downstream analyses, this information was also added to the alignment files in the ‘CB:Z’ tag. Cells with a total count of fragments below 5,000 were considered to have insufficient data and were filtered out. Single-cell chromatin mark analysis using ArchR The chromatin mark data we stratified into active marks (ATAC, H3K4me3, H3K27ac) and repressive marks (H3K9me3, H3K27me3) based on their impact on gene regulation. Fragment data of an individual mark was loaded and converted into Arrow files using the ‘createArrowFiles’ function of ArchR 47 v1.0.1. Filtering low-quality single-cell data was performed with the following parameters: minTSS = 1/0.1 (respectively for active/repressive marks), minFrags = 1000, addTileMat = TRUE, addGeneScoreMat = TRUE. We additionally identified and discarded potential doublets by using the ‘add DoubletScores’ function. All the Arrow files were load and consolidated into an ArchR project for analysis. Dimensionality reduction was performed by the ‘addIterativeLSI’ function with the following parameters: useMatrix = "TileMatrix", name = "IterativeLSI", iterations = 2, dimsToUse = 1:30. Clustering was performed using the ‘addClusters’ function with ‘method’ set as ‘Seurat’. An UMAP embedding was obtained by the ‘addUMAP’ function with default parameters. To annotate the cell identity of clusters, we firstly calculated gene score using the ‘addGeneScoreMatrix’ function on an individual mark. Here, the gene score serves as a proxy for the strength of the chromatin mark around the gene region. We calculated the marker genes of individual cell clusters by the ‘getMarkerFeatures’ function with the following parameters: useMatrix = "GeneScoreMatrix", groupBy = "Clusters", bias = c("TSSEnrichment", "log10(nFrags)"), testMethod = "wilcoxon". Marker genes were defined as those with log2-transformed fold change (log2FC) ≥ 1 and p-value ≤ 0.05. We annotated cell clusters by known cell type marker genes. Peak calling To reduce the effect of the sparsity of single-cell data on peak identification, fragment data were partitioned based on annotated cell types and further used for peak calling. Each fragment was extended by 75 bp on both sides to be biologically and computationally consistent with the short-read ATAC and CUT&TAG data. Extended fragments were used for peak calling. For ATAC data, we used the ‘callpeak’ command of MACS2 48 v2.2.9.1 with parameters “-f BED -q 0.05 --keep-dup all -B --SPMR”. For histone modification data, we used SEACR 49 v1.3 with parameters “non stringent 0.05”. Additionally, peaks supported by more than 2.5% of cells were considered reliable and retained for subsequent analysis. For bulk CUT&TAG-seq data, the parameters of peak calling were identical to those described above. Differential peak analysis of chromatin marks For parallel comparison, peaks from each cell type were consolidated into a unified list. Using the BEDtools ‘merge’ command, overlapping peaks were merged into single representative peaks while non-overlapping peaks remained unchanged. We then annotated whether each peak was identified in one or multiple cell types. The peak list was loaded into ArchR using the ‘addPeakSet’ function, and their signal strength in each cell was calculated by the ‘addPeakMatrix’ function. We calculated the differentially expressed peaks between two cell clusters by the ‘getMarkerFeatures’ function with the following parameters: useMatrix = "PeakMatrix", groupBy = "Clusters", bias = c("log10(nFrags)"), testMethod = "wilcoxon", useGroups = “cell_type1”, bgdGroups = “cell_type2”. The highly expressed peaks in cell type 1 required log2FC ≥ 1 and p-value ≤ 0.05, whereas those in cell type 2 required log2FC ≤ -1 and p-value ≤ 0.05. Function enrichment and motif analysis of peaks The functional GO analysis of the genes linked to marker or differential peaks was conducted by R package ClusterProfiler 50 v4.6.2 with a p-value cut-off of 0.01. Qualified GO terms with top gene ratios were visualized. In addition, similar GO terms were consolidated, with one representative term being kept. We extracted genomic regions spanning 1 kb upstream and downstream of TSS of genes. For active marks, we associated nearby peaks with corresponding genes using BEDtools ‘intersect’ command. The motif annotation of peaks was performed using the ‘addMotifAnnotations’ function with cisbp database. On the basis of the results of differential peak analysis, motif enrichment was performed using the ‘peakAnnoEnrichment’ function. To extract motifs enriched in cell type 1, the parameters were set as follows: peakAnnotation = "Motif", cutOff = "Pval = 0.5". To extract motifs enriched in cell type 2, the parameters were set as follows: peakAnnotation = "Motif", cutOff = "Pval < = 0.05 & Log2FC <= -0.5". Using the ‘getFeatures’ and ‘plotGroups’ function, the motif score of cells were extracted and visualized across cell types using boxplots. For visualization, we extracted the position weight matrix (PWM) of each motif using the ‘getPeakAnnotation’ in ArchR, which was further loaded into the R package ‘ggseqlogo’ 51 v0.2 to generate sequence logos. As an alternative approach, bed files containing peaks can also be used as input to the HOMER ( http://homer.ucsd.edu/homer/ ) software v5.1 for enriched motif calculation. Evaluation of the throughput of multi-modal data at single-cell level The four multi-modal data types comprised ATAC-seq (I7 adaptor) in combination with the following modifications: H3K4me3, H3K27ac, H3K9me3, and H3K27me3 (I5 adaptor). Those read ends without matched adaptors were named unclassified (UN). Firstly, we quantified the count, length, and total data size of sequencing reads per cell. After adaptor trimming, we calculated the count and proportion of the I5, I7 and UN fragments. Based on the adaptors, reads were classified into 5 types: I5_I5, I5_I7, I7_I7, UN_I5 and UN_I7. We quantified the count, proportion and length of modality-tagged reads. After read mapping, the count of uniquely mapped reads and fragments was quantified per cell. Then, the mapping ratio was defined as the ratio of aligned reads (MAPQ > 30) to the total number of modality-tagged reads. The total number of mapped bases per cell were calculated by the ‘depth’ command of SAMtools. The mapping coverage was defined as the number of mapped bases divided by the total size of the reference genome. For comparative analysis, we computed the statistics for in-house scNanoATAC-seq and scNanoSeq-CUT&Tag data in a manner consistent with the aforementioned methods. Evaluation of efficiency, reliability and reproducibility of scNanoCT&ATAC-seq We first evaluated individual modalities using the fractions of read ends in peaks (FRIP) and the transcriptional start site (TSS) enrichments. For FRIP, the peaks called from all cells of the modality served as the target regions. Using the ‘intersect’ command of BEDtools, we calculated the number of fragments falling within these peaks, which was then divided by the total number of fragments to obtain the FRIP. The TSS enrichments of each cell had already been calculated in previous quality control step of ArchR. The TSS distribution profile were extracted using the ‘plotTSSEnrichment’ function of ArchR. Subsequently, we evaluated the technical reproducibility across replicates. For each cell line and each set of multimodal data, we generated a comparable number of single-cell datasets in two separate experimental batches (196 cells per batch). We partitioned the genome into consecutive 10-kb bins. For each bin, we counted the number of tagged-fragments from each of the two batches. The data was structured into a matrix where rows represented bins, columns represented the two batches, and values contained the fragment counts. We then computed the correlation of fragment counts across all bins between the two batches using Pearson correlation method. At the peak level, the overlapped peaks between two batches were identified by BEDTools ‘intersect’ command, and the ratio of overlapped peaks to all peaks from individual batch were calculated as an indicator of reproducibility. For cross-platform validation, we used the peaks called by deep-sequencing bulk NGS-based data as a benchmark for evaluating the peak precision and recall. Using the ‘intersect’ command of BEDtools, we identified the scNanoCT&ATAC-seq peaks overlapped with NGS-based peaks as true-positive (TP) peaks. Meanwhile, the unique peaks of scNanoCT&ATAC-seq were defined as false-positive (FP) peaks, and the unique peaks of NGS data were defined as false-negative (FN) peaks. Then, we calculated the precision of peaks as TP/(TP + FP) and the recall of peaks as TP/(TP + FN). In addition to the global assessment, we further examined the precision and recall ratio of peaks by leveraging datasets of increasing cell numbers. Similarly, the correlation between bins and the consistence between peaks from the two techniques were evaluated as described previously. To visualize chromatin marks across the genome, we used deepTools 52 v3.5.5 to compute and plot signal distribution heatmaps spanning 1 kb/5 kb regions centered on peak summits. Genomic coverage tracks in bigwig format were generated with the ‘getGroupBW’ function in ArchR with parameters “tileSize = 100, maxCells = 10000”. For NGS-based data, we generated the BedGraph files using the BEDtools ‘genomecov’ command, then converted them to bigwig format using bedGraphToBigWig 53 v4. The bigwig files were loaded into Integrative Genomics Viewer (IGV) for visualization. Genomic characteristics and distribution of multimodal data It is widely acknowledged that the five modalities display characteristic genomic distributions, which also serves as an evaluation metric. Firstly, we evaluated the relative enrichment of individual modalities to NGS-based peaks as benchmark. In brief, we quantified the count of modality-tagged fragments overlapping the peaks. The counts were normalized by their corresponding library sizes and log2-transformed, generating a matrix where rows represent peaks and columns contain the relative enrichment scores for the five modalities. We then annotated the genomic location of peaks using the R package ChIPseeker 54 v1.34.1. We also annotated the candidate cis-regulatory elements (cCREs) 23 that covered by peaks using the BEDtools ‘intersect’ command. Structural variation (SV) detection and evaluation We performed SV calling on single-cell bam files with cuteSV 55 v1.0.10. The parameters were set as follow: --max_cluster_bias_INS 100 --diff_ratio_merging_INS 0.3 --max_cluster_bias_DEL 100 --diff_ratio_merging_DEL 0.3 --min_support 1. For each cell, any SV supported by at least one aligned read was retained and recorded in the vcf file. The five SV types were: deletions (DEL), insertions (INS), duplications (DUP), inversions (INV), and breakends (BND). For the first four SV types, we extracted genomic coordinates from vcf files and converted them into bed format. Within each cell type, all SVs were merged based on their type and genomic coordinates. We then quantified the number of supporting cells for each merged SV. The supporting cell count served as a reliability metric for downstream filtering. BND variants represent genomic rearrangements involving two distant genomic loci, with each VCF record containing paired coordinates. For integration, we binned these coordinates at 1000 bp resolution. BND events sharing identical binned coordinates were then consolidated. Subsequent quantification of supporting cells and filtering were performed consistent with the aforementioned methods. For benchmarking, we downloaded K562 bulk whole-genome data (WGS) generated by Nanopore sequencing from public sources 22 . Data were processed through the same mapping and SV calling pipeline, and only the SVs supported by more than 5 reads were retained to generate a high-confidence benchmark SV set. We identified the SVs of scNanoCT&ATAC-seq overlapped with SVs of WGS as true-positive (TP). Meanwhile, the unique SVs of scNanoCT&ATAC-seq were defined as false-positive (FP), and the unique SVs of WGS were defined as false-negative (FN). Then, we calculated the precision of SVs as TP/(TP + FP) and the recall of SVs as TP/(TP + FN). In addition to the global assessment, we further examined the precision and recall ratio of SVs supported by increasing cell numbers. CNV detection and evaluation We performed CNV quantification on single-cell bam files with Control-FREEC 56 v11.6b. The parameters were set as follow: ploidy = 2, breakPointThreshold = .8, window = 1000000, minExpectedGC = 0.30, maxExpectedGC = 0.60, sex = XY, inputFormat = BAM, mateOrientation = 0. The WGS data were processed through the same pipeline. We calculated the average CNV ratio of each 1 Mb window across all cells as a global metric, which was compared to the CNV ratio from WGS. The CNV ratio for each genomic window in every single cell was visualized as a heatmap, annotated with corresponding cell types and library sources. Assigning chromatin states from multimodal data at the bulk level To integrate information from multiple modalities, we used the multivariate HMM introduced in ChromHMM 25 v1.26. All the modality-tagged fragments from individual cells were merged to produce pseudo-bulk bed files. The configuration file documenting the cell type, modality, and file paths was processed by the ‘BinarizeBed’ command with default parameters. The resulting files binarized at 200 bp resolution were loaded into the ‘LearnModel’ function to model the combinatorial states and spatial patterns from five chromatin marks. The number of possible states were set as 10. Ten states were functionally annotated based on the high probability chromatin marks and the enrichment in specific genomic regions in state. We obtained the RNA-seq data of cell lines from ENCODE 23 to validate the biological relevance of ChromHMM-derived chromatin states. We extracted all genes within the 10 chromatin states, followed by a parallel comparison of expression levels across these states. As hypothesized, genes in active states exhibited significantly higher expression than those in repressive states. Assigning chromatin states from multimodal data at the single-cell level Single-cell chromatin state analysis was performed using scChromHMM, an extention of bulk ChromHMM framework established in prior work 15 , with specific adaptations for the scNanoCT&ATAC-seq dataset. The most critical step, termed “Anchor”, involves leveraging a reliable single-cell annotation set to associate multiple modalities of data with individual cells. Using the ATAC layer as a reliable reference, in this study, the modalities co-assayed in the same cell are intrinsically linked. For other modalities, we used the “FindTransferAnchors” algorithm in Seurat 26 v4 to identify anchor correspondences. Importantly, the number of modalities associated with individual cells is variable, with some cells encompassing two or more data types. ScChromHMM requires four input files. The pseudo-bulk HMM model parameters generated previously were used. Since scChromHMM is currently limited to processing 10x Genomics cell barcodes, we generated compatible pseudo-barcodes for each scNanoCT&ATAC-seq cell library. The fragment files, reference cell barcodes and anchors list were then generated according to the manual. We also maintained a cross-reference table between original and converted barcodes to ensure data provenance throughout the analysis. We sequentially executed the ‘hmm’ and ‘transform’ modules using default parameters. The output of scChromHMM is the posterior probabilities of each chromatin state for every reference cell at 200 bp resolution. In this study, scChromHMM analysis identified two primary chromatin states: Active and Repressed, serving as quantification metrics for comprehensive assessment of chromatin activity. Evaluation of scChromHMM application on scNanoCT&ATAC-seq For evaluation, we calculated the average active/repressed probabilities across all cells, counted and normalized the number of modality-tagged fragments of each 1 kb bin. The Pearson correlation between scChromHMM probabilities and individual modality signals were calculated. Then, we defined the A/R score as the ratio of active to repressed probabilities, and the regions with A/R score ≥ 2 were classified as active regions, while those with A/R score ≤ 0.5 were designated as repressed regions. We calculated the mean A/R scores by scChromHMM across 10 chromatin states of ChromHMM, demonstrating their concordance. To assess the potential of the A/R score as predictor for gene activity, we stratified genes based on their TPM values in ENCODE RNA-seq datasets into four categories: no expression (N/E, TPM = 0), low-level (Low, ≤ 25th percentile), medium-level (Mid, > 25th and ≤ 90th percentile), and high-level (High, > 90th percentile). We visualized the distribution of A/R scores across gene groups with boxplots and performed one-way ANOVA with Tukey's test for multiple comparisons. We next performed cell clustering based on the A/R score, an analogous process as single-modality analysis. For the ± 1 kb region surrounding TSS of each gene in each cell, we computed the A/R scores and integrated into a gene activity/expression matrix. The matrix were loaded and converted into a Seurat object using the ‘CreateSeuratObject’ function of Seurat with the parameters for quality control: min.cells = 5, min.features = 300. After choosing the most variable features and data scaling, we performed the Principal Component Analysis (PCA) for dimensionality reduction, cell clustering, and UMAP embedding with default parameters. We evaluated the concordance in cell clustering between the A/R score-based and ATAC-based approaches, and visualized the A/R score of cell line marker genes at the single-cell resolution. H3K9me3 enrichment on TEs Repetitive regions annotated by RepeatMasker were downloaded from UCSC 28 , 57 , from which we extracted five major classes of TEs (LINE, LTR, Retroposon, SINE, DNA) and their respective subcategories. We annotated the TEs that covered by peaks using the BEDtools ‘intersect’ command. We compared the enrichment of TEs in H3K9me3-modified versus unmodified regions through applying chi-square test to assess statistical significance and calculating the odds ratios. We also quantified the types, counts, and proportions of TEs overlapping with H3K9me3-marked regions. To evaluate the advantages of long-read sequencing in H3K9me3 profiling in TEs/repeats, we collected the bigwig files of three technologies: NGS-based ChIP-seq 23 , scNanoSeq-CUT&Tag, and scNanoCT&ATAC-seq. Using the ‘bigWigAverageOverBed’ command of UCSC 58 , we extracted the signal values for each TE and normalized them for parallel comparison. We counted the number of TEs uniquely and commonly identified across platforms, and assessed their strengths in TEs of increasing lengths. Categorization and characterization of H3K9me3 peaks In the chromatin state analysis, we identified not only the canonical H3K9me3 signals localized to heterochromatin (state E10) but also a novel category of H3K9me3 signals occurring at active transcription start sites (state E2). We characterized these two categories with respect to their peak length, correlation with gene expression, motif enrichment, TEs enrichment, and cCREs enrichment, following the same methods as described previously. We also downloaded the A/B compartment files from Rao et al. 59 , which is processed based on previous version of human reference genome (GRCh37). Therefore, we lifted the coordinates of H3K9me3 peaks to GRCh37 using ‘liftOver’ ( http://genome.ucsc.edu/cgi-bin/hgLiftOver ) and calculated their distribution on A/B compartments using the ‘intersect’ command of BEDtools. We further measured the co-localization between E2/E10 H3K9me3 peaks and ATAC peaks, and characterized the co-accessible chromatin marks at the opposite ends of H3K9me3-tagged reads, which served as indicators of roles in active transcription. For cross-validation, we intersected the scNanoCT&ATAC-seq H3K9me3 peaks separately with NGS-based ATAC-seq and H3K9me3 ChIP-seq peaks from ENCODE 23 . Identification of active promoters and enhancers We identified active promoters and enhancers across cell types by integrating chromatin states and genomic positions. Specifically, we extracted all ATAC/accessible peaks classified as E1 or E2 states as candidate promoters, and those in the E4 state as candidate enhancers. We applied more stringent criteria for promoter definition: 1) location within ± 1 kb of a TSS of transcripts; 2) when multiple peaks were present near the same TSS, only the peak closest to the TSS were selected; 3) any peak identified as an enhancer in at least one cell type retained the designation to ensure consistent promoter identity across cell types. After promoter annotation, the remaining peaks were classified as enhancers. To validate the reliability of definition, we found the overlap of our promoter/enhancer annotations with the cCREs from ENCODE using BEDtools. We further categorized the promoters/enhancers only detected in GM12878 or K562 as cell-specific and those with genomic coordinate overlaps as common. The statistics and pathway enrichment of the three peak categories were performed as aforementioned. Identification and characterization of E-P pairs Leveraging the unique characteristics of scNanoCT&ATAC-seq data: two ends of each read represent co-accessible loci and carry multiple chromatin marks. The number of read pairs spanning two peaks can serve as an indicator of peak co-accessibility. To ensure robustness, only E-P peak pairs supported by at least 10 ATAC/H3K4me3/H3K27ac read pairs were retained for subsequent analysis. Considering the library size selection and sequencing length limit of platform, we calculated the co-accessibility strength of E-P pairs across varied genomic distances, confirming its ability in predicting neighboring co-accessible peaks within the 1–10 kb range. For each cell line, we quantified and normalized the counts of read pairs in E-P pairs, which were sub-classified by cell line specificity, and visualized the clustered co-accessibility strength using the R package ComplexHeatmap 60 v2.14.0. To assess the potential of the E-P pairs as predictor for gene activity, we firstly annotated the heatmap with the difference in co-accessibility strength and the difference in RNA expression for each E-P pair of specific gene. Additionally, we used scatter plots to display their correlation by calculating the Pearson's correlation coefficient (R), associated p-value for significance, as well as the slope and intercept from the linear regression model. For the promoters associated with multiple regulatory enhancers, we grouped them by enhancer count and visualized the corresponding gene expression levels using boxplots. ANOVA test for assessing significant difference and a linear regression fitted to the mean TPM value of groups were performed to illustrate their correlation. Single-cell chromatin states analysis on LUAD cells The fundamental single-cell analysis followed the aforementioned procedures, with the following custom analyses. Cells were firstly clustered and annotated based on the ATAC profiles, identifying one cluster of normal epithelial cells (N) and two clusters of tumor cells (T1, T2). Bam files of each cell were used for inferring CNV by Control-FREEC, and fragment files from 3 cell clusters were used for peak calling. For each cell type, we broadly categorized H3K9me3 peaks overlapping with ATAC peaks into E2 states and other peaks into E10 states. We identified peaks gained in tumor using thresholds of log2FC ≥ 1 and p-value ≤ 0.05, and peaks loss in tumor with log2FC ≤ -1 and p-value ≤ 0.05. We assigned CNV ratios to peaks based on their location within 1 Mb genomic windows. By comparing the peak fold change (FC peak ) with the corresponding CNV ratio/fold change (CNV peak ), between tumor and normal cells, the crosstalk can be summarized into three classes: 1) de novo increases (FC peak - CNV peak ≥ 0.5); 2) de novo decreases (FC peak - CNV peak ≤ -0.5); and 3) passive increases (FC peak - CNV peak -0.5) on chromatin marks. For survival analysis, we collected the lung adenocarcinoma (LUAD) TCGA datasets from the UCSC Toil RNAseq Recompute Compendium datasets. We utilized the ‘survival’ 61 package v3.5.5 in R to classify patients into high-expression group (High) and low-expression group (Low) using the best expression cutoff. Then, the survival duration and status of patients in the two groups, Kaplan-Meier curves, log-rank test, and Cox proportional hazards regression on genes were performed to display clinical-related genes. Declarations Acknowledgments Our Flow Cytometry and Sequencing work was performed at the Advanced Cell Technology Core Facility, Guangzhou National Laboratory. This work was supported by grants from the Major Project of Guangzhou National Laboratory (GZNL2024A03001); Shenzhen Medical Research Fund (B2402019); Young Talents Program of Sun Yat-sen University Cancer Center (YTP-SYSUCC-0013); The National Key Research and Development Program Project (2024YFC3406203). Author contributions X.F., Z.Z. conceived the project and designed the experiments. D.S. collected the samples and performed library preparation experiments. Z.L. and E.D. did the bioinformatics work. L.L. and Q.H. did the FAM78A validation experiment. J.Z. did the cell culture work. W.X and Z.G. collected the LUAD sample. X. F., Z.Z., D.S. and Z.L. wrote the manuscript. All authors contributed to the discussion and interpretation of the results. Declaration of interests The authors declare no competing interests. Code availability Code for processing scNanoCT&ATAC-seq data and Illustrative code snippets for postprocessing are available at https://github.com/canceromics/scNanoCT-ATAC-seq . Data availability The raw sequencing data of the single cells have been deposited to Genome Sequence Archive (GSA) in the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences under accession code PRJCA05098. All other data are available in the article and its Supplementary files . References Soufi, A. et al. Pioneer Transcription Factors Target Partial DNA Motifs on Nucleosomes to Initiate Reprogramming. Cell 161, 555–568 (2015). Nicetto, D. et al. H3K9me3-heterochromatin loss at protein-coding genes enables developmental lineage specification. Science 363, 294–297 (2019). Xu, Z. et al. Structural variants drive context-dependent oncogene activation in cancer. Nature 612, 564–572 (2022). George, J. et al. Comprehensive genomic profiles of small cell lung cancer. Nature 524, 47–53 (2015). Karlić, R., Chung, H.-R., Lasserre, J., Vlahoviček, K. & Vingron, M. Histone modification levels are predictive for gene expression. Proc. Natl. Acad. Sci. U.S.A. 107, 2926–2931 (2010). Rao, S., Ahmad, K. & Ramachandran, S. Cooperative binding between distant transcription factors is a hallmark of active enhancers. Molecular Cell 81, 1651–1665.e4 (2021). Zhou, Z. et al. Composite transposons with bivalent histone marks function as RNA-dependent enhancers in cell fate regulation. Cell 188, 5878–5894.e18 (2025). Mas, G. et al. Promoter bivalency favors an open chromatin architecture in embryonic stem cells. Nat Genet 50, 1452–1462 (2018). Buenrostro, J. D. et al. Single-cell chromatin accessibility reveals principles of regulatory variation. Nature 523, 486–490 (2015). Cusanovich, D. A. et al. Multiplex single-cell profiling of chromatin accessibility by combinatorial cellular indexing. Science 348, 910–914 (2015). Kaya-Okur, H. S. et al. CUT&Tag for efficient epigenomic profiling of small samples and single cells. Nat Commun 10, 1930 (2019). Carter, B. et al. Mapping histone modifications in low cell number and single cells using antibody-guided chromatin tagmentation (ACT-seq). Nat Commun 10, 3747 (2019). Bartosovic, M., Kabbe, M. & Castelo-Branco, G. Single-cell CUT&Tag profiles histone modifications and transcription factors in complex tissues. Nat Biotechnol 39, 825–835 (2021). Wu, S. J. et al. Single-cell CUT&Tag analysis of chromatin modifications in differentiation and tumor progression. Nat Biotechnol 39, 819–824 (2021). Zhang, B. et al. Characterizing cellular heterogeneity in chromatin state with scCUT&Tag-pro. Nat Biotechnol 40, 1220–1230 (2022). Gopalan, S., Wang, Y., Harper, N. W., Garber, M. & Fazzio, T. G. Simultaneous profiling of multiple chromatin proteins in the same cells. Molecular Cell 81, 4736–4746.e5 (2021). Meers, M. P., Llagas, G., Janssens, D. H., Codomo, C. A. & Henikoff, S. Multifactorial profiling of epigenetic landscapes at single-cell resolution using MulTI-Tag. Nat Biotechnol 41, 708–716 (2023). Bartosovic, M. & Castelo-Branco, G. Multimodal chromatin profiling using nanobody-based single-cell CUT&Tag. Nat Biotechnol 41, 794–805 (2023). Stuart, T. et al. Nanobody-tethered transposition enables multifactorial chromatin profiling at single-cell resolution. Nat Biotechnol 41, 806–812 (2023). Tedesco, M. et al. Chromatin Velocity reveals epigenetic dynamics by single-cell profiling of heterochromatin and euchromatin. Nat Biotechnol 40, 235–244 (2022). Hu, Y. et al. scNanoATAC-seq: a long-read single-cell ATAC sequencing method to detect chromatin accessibility and genetic variants simultaneously within an individual cell. Cell Res 33, 83–86 (2022). Li, Q. et al. scNanoSeq-CUT&Tag: a single-cell long-read CUT&Tag sequencing method for efficient chromatin modification profiling within individual cells. Nat Methods 21, 2044–2057 (2024). The ENCODE Project Consortium et al. Perspectives on ENCODE. Nature 583, 693–698 (2020). Braun, T. P., Eide, C. A. & Druker, B. J. Response and Resistance to BCR-ABL1-Targeted Therapies. Cancer Cell 37, 530–542 (2020). Ernst, J. & Kellis, M. Chromatin-state discovery and genome annotation with ChromHMM. Nat Protoc 12, 2478–2492 (2017). Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587.e29 (2021). Chen, F., Kan, H. & Castranova, V. Methylation of Lysine 9 of Histone H3. in Handbook of Epigenetics 149–157 (Elsevier, 2011). doi: 10.1016/B978-0-12-375709-8.00010-1 . Hoyt, S. J. et al. From telomere to telomere: The transcriptional and epigenetic state of human repeat elements. Science 376, eabk3112 (2022). Pehrsson, E. C., Choudhary, M. N. K., Sundaram, V. & Wang, T. The epigenomic landscape of transposable elements across normal human development and anatomy. Nat Commun 10, 5640 (2019). Lee, E. et al. Landscape of Somatic Retrotransposition in Human Cancers. Science 337, 967–971 (2012). Xu, R. et al. Stage-specific H3K9me3 occupancy ensures retrotransposon silencing in human pre-implantation embryos. Cell Stem Cell 29, 1051–1066.e8 (2022). Lippman, Z. et al. Role of transposable elements in heterochromatin and epigenetic control. Nature 430, 471–476 (2004). Little, C. D., Nau, M. M., Carney, D. N., Gazdar, A. F. & Minna, J. D. Amplification and expression of the c-myc oncogene in human lung cancer cell lines. Nature 306, 194–196 (1983). Zhao, X. et al. Homozygous Deletions and Chromosome Amplifications in Human Lung Carcinomas Revealed by Single Nucleotide Polymorphism Array Analysis. Cancer Research 65, 5561–5570 (2005). Baykara, O., Bakir, B., Buyru, N., Kaynak, K. & Dalay, N. Amplification of Chromosome 8 Genes in Lung Cancer. J. Cancer 6, 270–275 (2015). Su, D., Peters, M., Soltys, V. & Chan, Y. F. Copy number normalization distinguishes differential signals driven by copy number differences in ATAC-seq and ChIP-seq. BMC Genomics 26, 306 (2025). Zhou, L. et al. ASB7 is a negative regulator of H3K9me3 homeostasis. Science 389, 309–316 (2025). Collins, R. L. & Talkowski, M. E. Diversity and consequences of structural variation in the human genome. Nat Rev Genet 26, 443–462 (2025). Xiong, H., Wang, Q., Li, C. C. & He, A. Single-cell joint profiling of multiple epigenetic proteins and gene transcription. Sci. Adv. 10, eadi3664 (2024). Corces, M. R. et al. An improved ATAC-seq protocol reduces background and enables interrogation of frozen tissues. Nat Methods 14, 959–962 (2017). Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet j. 17, 10 (2011). Li, H. Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics 34, 3094–3100 (2018). Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009). Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014). Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat Methods 9, 357–359 (2012). Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 (2010). Granja, J. M. et al. ArchR is a scalable software package for integrative single-cell chromatin accessibility analysis. Nat Genet 53, 403–411 (2021). Zhang, Y. et al. Model-based Analysis of ChIP-Seq (MACS). Genome Biol 9, R137 (2008). Meers, M. P., Tenenbaum, D. & Henikoff, S. Peak calling by Sparse Enrichment Analysis for CUT&RUN chromatin profiling. Epigenetics & Chromatin 12, 42 (2019). Xu, S. et al. Using clusterProfiler to characterize multiomics data. Nat Protoc 19, 3292–3320 (2024). Wagih, O. ggseqlogo: a versatile R package for drawing sequence logos. Bioinformatics 33, 3645–3647 (2017). Ramírez, F., Dündar, F., Diehl, S., Grüning, B. A. & Manke, T. deepTools: a flexible platform for exploring deep-sequencing data. Nucleic Acids Research 42, W187–W191 (2014). Kent, W. J., Zweig, A. S., Barber, G., Hinrichs, A. S. & Karolchik, D. BigWig and BigBed: enabling browsing of large distributed datasets. Bioinformatics 26, 2204–2207 (2010). Yu, G., Wang, L.-G. & He, Q.-Y. ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics 31, 2382–2383 (2015). Jiang, T. et al. Long-read-based human genomic structural variation detection with cuteSV. Genome Biol 21, 189 (2020). Boeva, V. et al. Control-FREEC: a tool for assessing copy number and allelic content using next-generation sequencing data. Bioinformatics 28, 423–425 (2012). Haeussler, M. et al. The UCSC Genome Browser database: 2019 update. Nucleic Acids Research 47, D853–D858 (2019). Pohl, A. & Beato, M. bwtool: a tool for bigWig files. Bioinformatics 30, 1618–1619 (2014). Rao, S. S. P. et al. A 3D Map of the Human Genome at Kilobase Resolution Reveals Principles of Chromatin Looping. Cell 159, 1665–1680 (2014). Gu, Z. Complex heatmap visualization. iMeta 1, e43 (2022). Therneau, T. M. & Grambsch, P. M. The Cox Model. in Modeling Survival Data: Extending the Cox Model 39–77 (Springer New York, New York, NY, 2000). doi: 10.1007/978-1-4757-3294-8_3 . Additional Declarations There is NO Competing Interest. Supplementary Files ExtendedDataTableS1.xlsx ExtendedDataTableS2.xlsx ExtendedDataTableS3.xlsx ExtendedFigure111.pdf ExtendedDataFig.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8125184","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":571192566,"identity":"8f4d53bd-2227-427b-a7ff-02d0d2a151d2","order_by":0,"name":"Xiaoying Fan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDUlEQVRIie2RsUoDQRCGZ1m4a8ak3aBcfISRBSEo56tsOLBKgmATCMQFIdVpHfAlfISTLdIotgemszTFlYu5wj1jYXOXKwX3gxmWZT/+HQbA4/mDhLrq5Iq7KgCOQOxRMPulsKW7aKn8wLGVEr5kHK/iiFYH5uP8c47dB80KC/GkVsGJYktKJJnO5dn4PkCxzngvheS6TrmAEUFBfPho8FSOU0TIFRwCZENdl9LdECi62SmDVGA/V3zbqIjvFFMp8h0sIeUqaE4RG3KzrGTPpbA7rfBk/bQYpJQ0fGxEHMtZ1Hl9loUt51H0dmtyO41rFUdoq36cQSDYojoyvVvuPvoaeAFli5cej8fz7/gCldFQZaj/6/0AAAAASUVORK5CYII=","orcid":"","institution":"Guangzhou National Laboratory","correspondingAuthor":true,"prefix":"","firstName":"Xiaoying","middleName":"","lastName":"Fan","suffix":""},{"id":571192567,"identity":"0c48927d-d889-4729-8ac0-80349dc84414","order_by":1,"name":"Dan Su","email":"","orcid":"","institution":"Guangzhou National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Dan","middleName":"","lastName":"Su","suffix":""},{"id":571192568,"identity":"022e67ea-6b52-47a7-a79a-71ec2e828584","order_by":2,"name":"Zhuobin Lin","email":"","orcid":"https://orcid.org/0000-0002-6076-5879","institution":"The Third Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Zhuobin","middleName":"","lastName":"Lin","suffix":""},{"id":571192569,"identity":"13de23ba-8497-4513-8eeb-6e7c9ec5ed4f","order_by":3,"name":"Lin Luo","email":"","orcid":"","institution":"Guangzhou National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Lin","middleName":"","lastName":"Luo","suffix":""},{"id":571192570,"identity":"ec8092cd-7f20-4cfe-8f43-70031aa20d15","order_by":4,"name":"Enze Deng","email":"","orcid":"","institution":"Guangzhou National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Enze","middleName":"","lastName":"Deng","suffix":""},{"id":571192571,"identity":"8e540865-4701-4b40-a13d-0d9c1ddaff81","order_by":5,"name":"Qingpei Huang","email":"","orcid":"","institution":"Guangzhou National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Qingpei","middleName":"","lastName":"Huang","suffix":""},{"id":571192572,"identity":"7f5cc02f-5148-4e1b-99f6-d96ce8dfbdf2","order_by":6,"name":"Wenxin Xu","email":"","orcid":"","institution":"Guangzhou National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Wenxin","middleName":"","lastName":"Xu","suffix":""},{"id":571192573,"identity":"2567a4b1-9052-4773-85d6-528aae3fffda","order_by":7,"name":"Zhihua Guo","email":"","orcid":"","institution":"the First Affiliated Hospital of Guangzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhihua","middleName":"","lastName":"Guo","suffix":""},{"id":571192574,"identity":"7962199a-a1b3-42b2-a0d5-d59122b4693a","order_by":8,"name":"Zhixiang Zuo","email":"","orcid":"","institution":"Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Zhixiang","middleName":"","lastName":"Zuo","suffix":""}],"badges":[],"createdAt":"2025-11-16 05:05:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8125184/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8125184/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100162934,"identity":"e8a82637-3ada-4c6e-bc12-5918548e640c","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":162893,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript2511v3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/3380c97acae6a106c50e2b2d.docx"},{"id":100367189,"identity":"c62963d3-4483-45e5-9bc0-6e25f5378724","added_by":"auto","created_at":"2026-01-16 07:56:49","extension":"json","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3135,"visible":true,"origin":"","legend":"","description":"","filename":"NMETHA63631.json","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/945f57fa8a896f4b852fc217.json"},{"id":100367883,"identity":"8cf7bf07-96f7-4de2-ae3b-189e2eb94e8f","added_by":"auto","created_at":"2026-01-16 07:57:25","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6531946,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedFigure111.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/020aa5b4dd0331293997c3fa.pdf"},{"id":100162938,"identity":"43b6dd26-a10d-431b-8f5a-6dc831068176","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"xml","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":194491,"visible":true,"origin":"","legend":"","description":"","filename":"NMETHA636310enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/09452ddf5dcb261074b1e230.xml"},{"id":100162942,"identity":"ce126e56-1bec-4199-ba1f-c740ae780fee","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"pdf","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4772259,"visible":true,"origin":"","legend":"","description":"","filename":"Figure16.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/18fe4d2611254435b6458606.pdf"},{"id":100369031,"identity":"1f3ba49f-f01e-45dd-8e4a-8db6d6de7229","added_by":"auto","created_at":"2026-01-16 07:58:38","extension":"xml","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":191539,"visible":true,"origin":"","legend":"","description":"","filename":"NMETHA636310structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/1a3ed35115bf2dafb25f651c.xml"},{"id":100162945,"identity":"f1a2b550-3c7d-48ad-808a-0f292334c509","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"html","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":219452,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/ba2cfbea9e8af3dba792d1d2.html"},{"id":100162929,"identity":"97b72956-0ad2-474d-b018-6ed43c66f248","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1117053,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eParallel profiling of multiple chromatin modalities in single cells by scNanoCT\u0026amp;ATAC-seq\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) The schematic diagram for the workflow of scNanoCT\u0026amp;ATAC-seq.\u003c/p\u003e\n\u003cp\u003e(b) Schematic of the library structures.\u003c/p\u003e\n\u003cp\u003e(c) Stacked bar chart showing the percentage of sequencing reads classified by fragment types with different read ends.\u003c/p\u003e\n\u003cp\u003e(d) Comparison of unique fragments per cell with SMS-based single modality methods.\u003c/p\u003e\n\u003cp\u003e(e) Comparison of the FRIP with SMS-based single modality methods.\u003c/p\u003e\n\u003cp\u003e(f) The top panel shows the averaged ATAC signal in peaks called in each reaction co-profiled with different histone modification. The bottom heatmap displays the ATAC signal of each peak in each reaction.\u003c/p\u003e\n\u003cp\u003e(g) The top panel shows the averaged Cut\u0026amp;Tag signal in peaks called in each reaction co-profiled with different histone modification. The bottom heatmap displays the Cut\u0026amp;Tag signal of each peak in each reaction.\u003c/p\u003e\n\u003cp\u003e(h) Presentative IGV track view showing the signals of ATAC and four histone marks of GM12878 cells from scNanoCT\u0026amp;ATAC-seq and ENCODE ChIP-seq data.\u003c/p\u003e\n\u003cp\u003ePrecision (left) and recall (right) of peak detection of the four histone modifications comparing to the bulk data.\u003c/p\u003e","description":"","filename":"fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/63289751e3ff6635c31ee296.jpg"},{"id":100368759,"identity":"56ab5b82-0573-4132-8ea9-e9ef0062572e","added_by":"auto","created_at":"2026-01-16 07:58:19","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1138350,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eScNanoCT\u0026amp;ATAC-seq accurately captures genomic and epigenetic differences between different cell types\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) UMAP visualization of the single cells based on ATAC and the four histone modifications, respectively.\u003c/p\u003e\n\u003cp\u003e(b) Heatmap displaying the differential peaks of each modality between K562 and GM12878 cells. The peak-related representative marker genes are indicated.\u003c/p\u003e\n\u003cp\u003e(c, d) IGV tracks showing the signals of ATAC and four histone marks in the marker gene region of each cell type. BCR is a maker for K562 cells and MS4A1 is maker for GM12878.\u003c/p\u003e\n\u003cp\u003e(e) Performance of SV identification in K562 cells in different reactions.\u003c/p\u003e\n\u003cp\u003e(f) Scatter plot showing the breakpoint events ranked by the number of supporting single cells.\u003c/p\u003e\n\u003cp\u003e(g) Heatmap showing the CNV landscape of single GM12878 and K562 cells.\u003c/p\u003e","description":"","filename":"fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/b810aa623012c471b594e0dc.jpg"},{"id":100162933,"identity":"8b5be2dd-1406-4426-955d-dab717e73110","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":791430,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFunctional genomic characterization with combinatorial chromatin modalities in the same cells\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Ten chromatin states defined by ChromHMM using five epigenetic signatures profiled by scNanoCT\u0026amp;ATAC-seq in GM12878 and K562 cells.\u003c/p\u003e\n\u003cp\u003e(b) Average expression levels of genes (top) and the proportion of expressed genes (bottom) located in each chromatin state in GM12878 cells.\u003c/p\u003e\n\u003cp\u003e(c) Average signal of each epi-modality in the active (E1-E5, top) and repressive (E6-E10, bottom) regions in GM12878 cells.\u003c/p\u003e\n\u003cp\u003e(d) Heatmap showing the Pearson correlation coefficients between the signal of each epi-modality and the active/repressive scores calculated by scChromHMM.\u003c/p\u003e\n\u003cp\u003e(e) Heatmap of the active-to-repressive (A/R) score for the ten chromatin states in GM12878 and K562 cells.\u003c/p\u003e\n\u003cp\u003e(f) Box plots showing the distribution of A/R scores for genes grouped by their expression level within each cell type. Genes are divided into four categories: not expressed (N.E.), low, medium (Mid), and high expression. The statistical test was done by Kruskal-Wallis test, p \u0026lt; 2.2e-16.\u003c/p\u003e\n\u003cp\u003e(a) UMAP visualization of the single cells based on A/R score.\u003c/p\u003e\n\u003cp\u003e(g) Heatmaps displaying the single-cell A/R scores for presentative cell type marker genes in GM12878 and K562 cells. Dot plots showing the mean A/R scores and the number of cells detected the gene in each cell type.\u003c/p\u003e\n\u003cp\u003e(h) IGV tracks showing the A/R overlay signals and the signals of ATAC and four histone marks in the representative gene regions.\u003c/p\u003e","description":"","filename":"fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/e1495ff2e582f0db7c7f9fa8.jpg"},{"id":100369576,"identity":"c0f46e57-52ba-4ad7-83fc-337756925002","added_by":"auto","created_at":"2026-01-16 07:59:09","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":981773,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCharacters of the two modes of H3K9me3 modification.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Schematic depiction of two distinct modes of H3K9me3 modification.\u003c/p\u003e\n\u003cp\u003e(b) Distribution of the peak length for the two modes of H3K9me3 peaks in GM12878 cells.\u003c/p\u003e\n\u003cp\u003e(c) Relative enrichment of H3K9me3 to ATAC signal in each mode of H3K9me3 peaks in GM12878 cells.\u003c/p\u003e\n\u003cp\u003e(d) Average expression levels of genes (left) and the proportion of expressed genes (right) located in each mode of H3K9me3 peaks in GM12878 cells.\u003c/p\u003e\n\u003cp\u003e(e) TF motif enrichment of ATAC, E2-H3K9me3 and E10-H3K9me3 peaks in GM12878 (top) and K562 (bottom) cells. The top five TF motifs are displayed with corresponding motif logos.\u003c/p\u003e\n\u003cp\u003e(f) \u0026nbsp;Venn diagram showing the overlap of the top 30 TFs enriched in ATAC, E2-H3K9me3 and E10-H3K9me3 peaks in GM12878 (top) and K562 (bottom) cells.\u003c/p\u003e\n\u003cp\u003e(g) The number of TE copies detected by each method (top) and the averaged H3K9me3 signal for the TE copies with different lengths (bottom).\u003c/p\u003e\n\u003cp\u003e(h) IGV tracks showing the coverage on the TE copies (shadowed regions) by different methods in each cell type.\u003c/p\u003e\n\u003cp\u003e(i) Dotplot showing the enrichment of each mode of H3K9me3 peaks in the TE families.\u003c/p\u003e","description":"","filename":"fig4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/c479ba8dd2422f1476324659.jpg"},{"id":100162940,"identity":"8a2278a3-aa2e-4015-b25e-632ab12e59d7","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":745374,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEvaluating E-P interaction at single-molecular level\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Schematic depiction how E-P interactions being identified and measured with scNanoCT\u0026amp;ATAC-seq data.\u003c/p\u003e\n\u003cp\u003e(b) Venn diagrams showing the overlap of promoters, enhancers and E-P pairs between GM12878 and K562 cells.\u003c/p\u003e\n\u003cp\u003e(c) Representative GO terms of cell type specific promoters.\u003c/p\u003e\n\u003cp\u003e(d) Heatmap showing the interaction scores of common and cell type specific enhancers and promoters. Relative expression is calculated as Log2Foldchange of TPM values between GM12878 and K562.\u003c/p\u003e\n\u003cp\u003e(e) Dotplot showing the interaction strength of each type of E-P pairs within each cell type.\u003c/p\u003e\n\u003cp\u003e(f) IGV tracks showing the reads supporting different enhancers interacting with KDM8 promoter in GM12878 and K562 cells respectively. Active CREs (aCREs) are defined by our data.\u003c/p\u003e\n\u003cp\u003e(g) IGV tracks showing reads supporting the new E (E15695, marked by yellow shadow)-P (P15568, marked by red shadow) pair formed through fusion of chr13 and chr9 regions.\u003c/p\u003e\n\u003cp\u003e(h) Schematic depiction how the new E-P (FAM78A) pair formed through translocation events and the validation procedure.\u003c/p\u003e\n\u003cp\u003e(i) Barplot showing the expression changes of FAM78A and the negative control gene RAPGEF1 after enhancer deletion by Cas9. sg13 and sg24 are two pairs of sgRNAs used to target the candidate enhancer region. The statistical test was done by unpaired t-tests. **p\u0026lt;0.01; ns: non-significant.\u003c/p\u003e","description":"","filename":"fig5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/35bba614a974039dba128fb1.jpg"},{"id":100367855,"identity":"c3bb37f7-4888-4d04-99f5-bea1e392db71","added_by":"auto","created_at":"2026-01-16 07:57:24","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1057403,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGenetic and epigenetic changes of tumor cells in LUAD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a)\u0026nbsp; Schematic diagram of LUAD sample processing.\u003c/p\u003e\n\u003cp\u003e(b)\u0026nbsp; UMAP showing the distribution of cells based on ATAC and H3K9me3 signals respectively. Cells with pair-wised modalities are connected by lines.\u003c/p\u003e\n\u003cp\u003e(c)\u0026nbsp; Dotplot showing the normalized gene scores of the classical marker genes in each cell type.\u003c/p\u003e\n\u003cp\u003e(d)\u0026nbsp; Heatmap of the genome-wide CNV patterns (excluding sex chromosomes) of each single cell. The copy ratios are calculated in 1Mb window. The CNVs in the chr8 are zoomed in in the bottom. The green dashed box indicates the region with copy number gaining in tumor_2 cells and the blue dashed box indicates the region with copy number gaining in tumor_1 cells. Representative tumor genes in each region are indicated in the bottom.\u003c/p\u003e\n\u003cp\u003e(e)\u0026nbsp; IGV tracks showing the ATAC signals of the representative genes in the normal and tumor cells.\u003c/p\u003e\n\u003cp\u003e(f)\u0026nbsp;\u0026nbsp;\u0026nbsp; Schematic depiction the three types of epigenetic changes under copy number increase.\u003c/p\u003e\n\u003cp\u003e(g)\u0026nbsp; Comparison of the fold changes of the ATAC signals and genomic copy numbers between normal and tumor_1 cells in the ATAC peak regions. \u003cem\u003eDe novo\u003c/em\u003e increase: FC\u003csub\u003eATAC\u003c/sub\u003e - FC\u003csub\u003eCNV\u003c/sub\u003e ≥ 0.5; \u003cem\u003eDe novo\u003c/em\u003e decrease: FC\u003csub\u003eATAC\u003c/sub\u003e - FC\u003csub\u003eCNV\u003c/sub\u003e ≤ -0.5); Passive increase: FC\u003csub\u003eATAC\u003c/sub\u003e - FC\u003csub\u003eCNV\u003c/sub\u003e \u0026lt; 0.5 and FC\u003csub\u003eATAC\u003c/sub\u003e - FC\u003csub\u003eCNV\u003c/sub\u003e \u0026gt; -0.5. The representative related genes are labeled.\u003c/p\u003e\n\u003cp\u003e(h)\u0026nbsp; Enriched GO terms for the peak-related genes in each category.\u003c/p\u003e","description":"","filename":"fig6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/10c87cc37a52c9e4e5d77d5f.jpg"},{"id":106094098,"identity":"9cc84126-b73d-4b29-9993-a0a0ef38748c","added_by":"auto","created_at":"2026-04-03 11:40:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":7538367,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/bd65972c-4c46-4a3a-8059-84795c8eddf5.pdf"},{"id":100162931,"identity":"fef8eced-9798-4896-b1ad-57a0586cac90","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":713617,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"ExtendedDataTableS1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/21c0b541d0289f4bb3a0bd50.xlsx"},{"id":100162937,"identity":"5c78ad49-db20-4467-bf4c-4f299c5b66ae","added_by":"auto","created_at":"2026-01-13 15:10:13","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":2711418,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"ExtendedDataTableS2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/8faf5377bc374be2baba8d35.xlsx"},{"id":100368341,"identity":"a2a9ca3e-dd29-4bd1-b623-23876372b9b4","added_by":"auto","created_at":"2026-01-16 07:57:52","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":466790,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"ExtendedDataTableS3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/e0f9368e181443caa7c09b8a.xlsx"},{"id":100367884,"identity":"a6281815-538f-4ab0-8885-37b746f7d519","added_by":"auto","created_at":"2026-01-16 07:57:25","extension":"pdf","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":6531946,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"ExtendedFigure111.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/683a8e2f2c137db9f3f4eb2e.pdf"},{"id":100367617,"identity":"7d8f42d1-70dc-41a4-960f-54ff84d9f56a","added_by":"auto","created_at":"2026-01-16 07:57:10","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":19529,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDataFig.docx","url":"https://assets-eu.researchsquare.com/files/rs-8125184/v1/2d20e94722d93b17df22b122.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Parallel profiling of genome mutations and multiple chromatin modalities in single cells by scNanoCT\u0026ATAC-seq","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe genome and epigenome, encompassing transcription factor (TF) binding, histone modifications and chromatin remodeling, are orchestrated to govern cell fate transitions\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Together, these factors determine the regulatory logic underlying cell state changes during development and disease progression. Alterations in genomic structures and chromatin states often precede transcriptional changes\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Foundational principles of chromatin regulatory include the cooperative binding of epigenetic factors, the co-occurrence of synergistic or antagonistic histone modifications, and the dynamic shifts in chromatin accessibility. For example, cis-regulatory elements are typically marked by the presence of multiple active histone marks and high chromatin accessibility, whereas bivalent loci are characterized by the presence of both active and repressive histone marks alongside constrained accessibility\u003csup\u003e\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Such epistatic interactions across epigenetic modalities are crucial for the rapid activation or repression of gene expression. Therefore, comprehensive multimodal chromatin mapping holds promise for predicting cell lineage commitment or cellular state transition prior to overt changes in transcriptional programs.\u003c/p\u003e \u003cp\u003eRecent advances in single-cell epigenomic technologies, such as single-cell assay for transposase-accessible chromatin with sequencing (scATAC-seq)\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, single-cell CUT\u0026amp;Tag (scCUT\u0026amp;Tag)\u003csup\u003e\u003cspan additionalcitationids=\"CR12 CR13 CR14\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, and their multi-modal derivatives have greatly enhanced the resolution of chromatin profiling. Multi-CUT\u0026amp;Tag, for example, enables simultaneously profiling of two chromatin proteins within the same cells by pre-coupling antibodies to pA-Tn5 loaded with distinct barcoded adapters\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Further increasing multiplexing capability, scNano-CUT\u0026amp;Tag employs nanobody-Tn5 fusions to profile up to three epigenetic modalities at single-cell resolution\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. However, its reliance on customized nano-antibodies limits broad adoption. Although these methods effectively map euchromatic marks such as H3K27ac and H3K27me3, their performance in capturing heterochromatic regions marked by H3K9me3 remains unclear. Among published single-cell methods, scGET-seq\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e has been shown to probe both heterochromatin and euchromatin by utilizing an engineered transposase fused to the chromodomain of the heterochrmatin protein-1a (HP-1a), offering a more detailed description of genomic copy number variations (CNVs) compared to scATAC-seq.\u0026nbsp;Still, none of these methods can capture structure variations (SVs) and epigenetic modalities in the same individual cells.\u003c/p\u003e \u003cp\u003eThe combination of long-read sequencing with single-cell assays has opened new avenues for studying genetic regulation. For instance, scNanoATAC-seq\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e combines scATAC-seq with Nanopore sequencing to simultaneously assess chromatin accessibility and genomic SVs. More recently, the same team developed scNanoSeq-CUT\u0026amp;Tag\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, which enables histone mark profiling in complex genomic regions that are challenging to study with short reads. Nevertheless, the potential of long-read sequencing for simultaneous profiling of histone modifications and chromatin accessibility has yet to be fully explored.\u003c/p\u003e \u003cp\u003eIn this study, we present scNanoCT\u0026amp;ATAC-seq, a single-cell assay that leverages Nanopore sequencing to simultaneously profile genome variations, histone modifications, and chromatin accessibility within individual cells. Our approach demonstrates comparable sensitivity and accuracy with established single-modality methods, including scNanoATAC-seq and scNanoSeq-CUT\u0026amp;Tag. We show, at single-molecule resolution, that genomic SVs can induce novel enhancer\u0026ndash;promoter (E-P) interactions in cancer cells. Applied to a lung adenocarcinoma (LUAD) sample, scNanoCT\u0026amp;ATAC-seq clearly separates tumor and normal epithelial cells based on both genomic and epigenomic features, and further resolves intratumoral subclones. We anticipate that scNanoCT\u0026amp;ATAC-seq will represent a valuable addition to the toolkit for multimodal single-cell profiling technologies.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003escNanoCT\u0026amp;ATAC-seq efficiently maps chromatin accessibility and histone modification landmarks in the same individual cells\u003c/h2\u003e \u003cp\u003eWe coupled protein A/G-fused Tn5 (pA/G-Tn5) and Tn5 with different adaptor sequences, referred to as I5-pA/G-Tn5 and I7-Tn5, respectively. In the scNanoCT\u0026amp;ATAC-seq approach (Fig.\u0026nbsp;1a), single-cell suspensions are incubated with primary antibodies targeting specific histone modifications or TFs. Afterward, the cells are treated with a secondary antibody and I5-pA/G-Tn5. Following washing out the free I5-pA/G-Tn5, the cells are incubated with I7-Tn5 while Mg\u003csup\u003e2+\u003c/sup\u003e ions are added to activate the tagmentation reactions. Subsequently, individual tagmentated cells are sorted into 96-well plates containing cell lysis buffer and preamplification primers, with each well containing two primer sequences sharing the same 16-bp inner cell barcode. After protein lysis, the DNA fragments undergo a first round of PCR to generate long amplicons. The amplicons with distinct inner cell barcodes are pooled and undergo a second round of PCR to tag them with outer cell barcodes. Ultimately, 600\u0026ndash;1000 cell amplicons containing diverse combinational barcodes are pooled together for library construction and sequencing on the Oxford Nanopore platform (Fig.\u0026nbsp;1a). The median fragment length of scNanoCT\u0026amp;ATAC-seq libraries ranges from 3,528 bp to 4,149 bp (Extended Data Fig.\u0026nbsp;1a,b).\u003c/p\u003e \u003cp\u003eAccording to the library structure, the histone marks and the open chromatin regions are recognized by I5 ends and I7 ends of the reads, respectively (Fig.\u0026nbsp;1b). We summarized the percentages of reads with different ends for each cell. In GM12878 cells, approximately 99% of reads were usable, as either end could be recognized by a specific adaptor (Fig.\u0026nbsp;1c). Moreover, the ratio of I5 to I7 ends varies across combinations with different histone marks. For example, I5-containing reads accounted for only 51.69% in H3K9me3 scNanoCT\u0026amp;ATAC-seq libraries, while they constituted 77.97% in H3K4me3 scNanoCT\u0026amp;ATAC-seq libraries. Reads with heterogenous ends were shorter, indicating crosstalk between open chromatin and histone modifications across the genome (Extended Data Fig.\u0026nbsp;1c). Compared to long-read sequencing-based single-modality methods, including scNanoATAC-seq\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e and scNanoSeq-CUT\u0026amp;Tag\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, our multiomics approach achieves a comparable mapping ratio (~\u0026thinsp;98%, except for the H3K9me3 group, Extended Data Fig.\u0026nbsp;1d) and sensitivity (Fig.\u0026nbsp;1d, up to 41,775 and 39,134 unique reads per cell for ATAC and histone modifications, respectively). With 2,700 to 417,504 reads per cell, the median single-cell genome coverage for scNanoCT\u0026amp;ATAC-seq is 5.84% (Extended Data Fig.\u0026nbsp;1e). Furthermore, transcription start site (TSS) scores from ATAC modality ranged from 1.02 to 11.70, higher than those from the H3K4me3 modality, while the H3K9me3 modality showed the lowest TSS scores (Extended Data Fig.\u0026nbsp;1f-g). Specifically, the ATAC signal from scNanoCT\u0026amp;ATAC-seq displayed nucleosome occupancy pattern around TSS, similar to that observed in scNanoATAC-seq (Extended Data Fig.\u0026nbsp;1h). The fraction of reads in peaks (FRIP) for each modality from scNanoCT\u0026amp;ATAC-seq data was also comparable to corresponding single-modality methods (Fig.\u0026nbsp;1e). These results suggest scNanoCT\u0026amp;ATAC-seq is highly sensitively in detecting both histone marks and open chromatin regions, with low signal-to-noise ratios, similar to previous single-omics methods.\u003c/p\u003e \u003cp\u003eNext, we evaluated the accuracy and reproducibility of scNanoCT\u0026amp;ATAC-seq in profiling multiple chromatin modalities. Libraries from different batches exhibited high consistency within each modality (Extended Data Fig.\u0026nbsp;2a). The ATAC module demonstrated remarkable similarity across different combinations of histone marks, while H3K4me3, H3K27ac, H3K27me3 and H3K9me3 exhibited distinct patterns (Fig.\u0026nbsp;1f-g, Extended Data Fig.\u0026nbsp;2b-c). We merged cells from all replicates (a total of ~\u0026thinsp;400 cells per histone mark, Extended Data Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and extracted peaks for each chromatin modality, which largely overlapped with the corresponding bulk profiles from the Encyclopedia of DNA Elements (ENCODE)\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e (Extended Data Fig.\u0026nbsp;2c-d, see methods). Each chromatin modality demonstrated significant positive enrichment in its corresponding bulk annotated region, with H3K9me3 largely excluded from active regions (Extended Data Fig.\u0026nbsp;3a). Genome browser tracks of merged scNanoCT\u0026amp;ATAC-seq cells indicated that each epigenetic modality closely resembled the ENCODE references for GM12878, while distinct modalities within the same population were markedly different, particularly for active and repressive marks (Fig.\u0026nbsp;1h). The precision of peak calling was globally higher for active histone marks, remaining stable (over 90%) when the sample size exceeded 50 cells (Fig.\u0026nbsp;1i). Additionally, around 50% of H3K4me3 and H3K27me3 bulk peaks were recaptured in our datasets (Fig.\u0026nbsp;1i).\u003c/p\u003e \u003cp\u003eFurther profile annotation revealed that over 50% H3K4me3 peaks corresponded to promoters, whereas H3K9me3-modified regions were predominantly located in distal intergenic regions (Extended Data Fig.\u0026nbsp;3b). Similarly, over 60% of unique ATAC, H3K4me3 and H3K27ac peaks identified from scNanoCTA-seq data were enriched at ENCODE candidate cis-regulatory elements (cCREs), while only 20% of H3K9me3 peaks were recognized as cCREs (Extended Data Fig.\u0026nbsp;3c), indicating efficient and accurate capture of peaks associated with active regulatory elements. Collectively, our analysis demonstrates that scNanoCT\u0026amp;ATAC-seq can efficiently map chromatin accessibility and histone mark occupancy in single cells.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eScNanoCT\u0026ATAC-seq accurately captures genomic and epigenetic differences between different cell types\u003c/h3\u003e\n\u003cp\u003eNext, we examined the ability of scNanoCT\u0026amp;ATAC-seq to distinguish between different cell types. Unsupervised clustering further revealed that the GM12878 and K562 cells were clearly separated in the Uniform manifold approximation and projection (UMAP) analysis for each modality (Fig.\u0026nbsp;2a, Extended Data Fig.\u0026nbsp;4a-b). The two types of cells exhibited a low abundance of common peaks for each epigenetic modality (Extended Data Fig.\u0026nbsp;4c). The differentially accessible regions (DARs) identified in GM12878 and K562 cells were enriched in pathways related to the B cells and leukocytes, respectively, underscoring the accuracy of scNanoCT\u0026amp;ATAC-seq in capturing the epigenetic characters of individual cell types (Extended Data Fig.\u0026nbsp;4d). B cell markers such as MS4A1, CD79A and HLA-DRB1 displayed higher intensities of ATAC, H3K4me3 and H3K27ac signals in GM12878 cells, whereas K562 cells showed elevated peak intensities in leukocyte-related genes including BCR, ABL1, GATA1, FAM30A (Fig.\u0026nbsp;2b-d, Extended Data Fig.\u0026nbsp;4e). Interestingly, these markers also exhibited scattered signals for repressive marks H3K27me3 and H3K9me3 in their corresponding cell types, suggesting that highly active regions are also dynamically modified by repressive histone marks (Fig.\u0026nbsp;2b-d).\u003c/p\u003e \u003cp\u003eSince scNanoCT\u0026amp;ATAC-seq also covers normal genome fragments by the long reads (Fig.\u0026nbsp;1b), we evaluated its capacity to detect SVs\u0026ndash;including insertions, deletions, inversions and duplications\u0026ndash;in K562 cells. Compared to ONT bulk sequencing data\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e, scNanoCT\u0026amp;ATAC-seq data effectively restored the identified control SVs, within highly consistent breakpoints (Extended Data Fig.\u0026nbsp;5a-b and Extended Data Table S2). The precision of SV detection increased to over 75% when supported by just three cells (Fig.\u0026nbsp;2e). Using fewer than 400 cells (Extended Data Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e), nearly 90% of the control SVs were recalled, with 3\u0026ndash;4 supporting cells being the recommended criteria to achieve both high precision and recovery for SVs (Fig.\u0026nbsp;2e). Notably, the classical BCR-ABL1\u003csup\u003e24\u003c/sup\u003e fusion ranked in the top two among all identified K562 variants (Fig.\u0026nbsp;2f, Extended Data Table S2). We also observed that the most frequent SV occurred in intergenic regions (chr13: 108,009,126 - chr9:FAM78A), which was also detected in the control and being validated unique in K562 cells (Fig.\u0026nbsp;2f, Extended Data Fig.\u0026nbsp;5c-d). With superior genome coverage over short-read sequencing based singe-cell epigenome sequencing methods, scNanoCT\u0026amp;ATAC-seq revealed a substantial abundance of CNVs unique to K562 cells, distinct from those in GM12878 cells (Fig.\u0026nbsp;2g, Extended Data Fig.\u0026nbsp;6). These collective analyses demonstrate that scNanoCT\u0026amp;ATAC-seq efficiently captures specific genomic and epigenomic characteristics in single cells.\u003c/p\u003e\n\u003ch3\u003eFunctional genomic characterization with combinatorial chromatin modalities in the same cells\u003c/h3\u003e\n\u003cp\u003eAs scNanoCT\u0026amp;ATAC-seq generates ATAC signal for every cell together with one of the four histone marks in the current dataset, we directly integrated all the single-cell profiles using ATAC modality (Fig.\u0026nbsp;2a). Then we assigned the chromatin states by ChromHMM\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e at pseudobulk level. A total of 10 chromatin states were obtained by running the Baum-Welch step of ChromHMM (Fig.\u0026nbsp;3a, Extended Data Fig.\u0026nbsp;7a-c), which grouped into: E1, flanking TSS (strongly enriched for H3K4me3 and weekly for H3K27ac/ATAC); E2, active TSS (strongly enriched for H3K4me3/H3K27ac/ATAC); E3, strong transcription (enriched in none of the epi-modalities but around TSS); E4, genic enhancer (enriched for H3K27ac/ATAC); E5, unmarked TSS (strongly enriched for only ATAC); E6, bivalent TSS (enriched for H3K27me3/ATAC); E7, weak repressed polycomb (weekly enriched for H3K27me3); E8, Repressed polycomb (enriched for H3K27me3 and low ATAC); E9, quiescent states (with no epigenetic signals detected); E10, heterochromatic states (enriched for H3K9me3). The distribution of most genomic elements slightly changed between GM12878 and K562 cells, especially for the E4 genic enhancers, which showed elevated participation in genic region in K562 cells (Extended Data Fig.\u0026nbsp;7a-c).\u003c/p\u003e \u003cp\u003eIn consistent to the element definition, E1 to E5 contained higher proportions of expressing genes and the general expression levels were higher. On the contrary, E6 to E10 genes showed hardly expressions (Fig.\u0026nbsp;3b, Extended Data Fig.\u0026nbsp;7d). Thus we categorized E1 to E5 as active chromatin regions, and E6 to E10 as repressive chromatin regions. The active histone marks, H3K4me3 and H3K27ac, showed much higher signals than the repressive histone mark H3K27me3 in the active regions. While H3K27me3 signal was much higher than H3K4me3 and H3K27ac signals in the repressive regions (Fig.\u0026nbsp;3c, Extended Data Fig.\u0026nbsp;7e). These results indicate that the comprehensive use of the five epigenetic modalities is accurate in defining the chromatin states.\u003c/p\u003e \u003cp\u003ePrevious study developed scChromHMM\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, which assigns the chromatin states at single-cell resolution with bioinformatically interpolated multiple histone marks. We tested the performance of scChromHMM in our scNanoCT\u0026amp;ATAC-seq data. The analysis returned the active and repressive probability for each 200 bp window across the whole genome in every single cell (see methods). The averaged active scores were postively correlated with ATAC, H3K4me3 and H3K27ac signals, while the averaged repressive scores were positively correlated with H3K27me3 signal (Fig.\u0026nbsp;3d). As large amount of genomic regions contains both active and repressive marks, we used active to repressive (A/R) score to measure the final transcriptional activity of each element. The bulk level defined active regions, E1 to E5, all showed larger A/R scores, while E6 to E10 had low A/R scores (Fig.\u0026nbsp;3e). We further divided genes to four groups based on their expression levels within each cell type, and compared the A/R scores of each group of genes. The A/R scores were significantly elevated with increased gene expressions. These data suggests that scNanoCT\u0026amp;ATAC-seq allows accurately measuring the chromatin transcription activity at single cell level.\u003c/p\u003e \u003cp\u003eWe further evaluated whether the A/R score could capture cell type specific transcriptional regulations by unsupervised clustering of all the single cells. The GM12878 and K562 cells could be clearly separated by A/R score, without batch effects from different modalities (Fig.\u0026nbsp;3g, Extended Data Fig.\u0026nbsp;7f). The classical markers of the two cell types, such as MS4A1, HLA-DRB1 for GM12878 and BCR, HBG2 for K562, showed higher A/R scores in single cells of the corresponding populations (Fig.\u0026nbsp;3h, Extended Data Fig.\u0026nbsp;7g). IGV track of the cell-type specific genes revealed higher intensity of both active and repressive signals in the corresponding cells, but the former was higher (Fig.\u0026nbsp;3i).\u003c/p\u003e \u003cp\u003e \u003cb\u003eTwo modes of H3K9me3 were identified.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eH3K9me3 is regarded as a landmark of repeat-rich constitutive heterochromatin\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Interestingly, according to the chromatin states annotation, H3K9me3 shows high priority in both E2 and E10 states (Fig.\u0026nbsp;3a, Extended Data Fig.\u0026nbsp;8a), which are annotated as active TSS and heterochromatin, respectively. These two categories of peaks represents two different modes of H3K9me3 modification (Fig.\u0026nbsp;4a). In both GM12878 and K562 cells, the E2-H3K9me3 peaks are short, highly accessible, gene rich and transcriptional active, while E10-H3K9me3 peaks are typical heterochromatin representative (Fig.\u0026nbsp;4b-d, Extended Data Fig.\u0026nbsp;8b-d). These ATAC peak overlapped H3K9me3 peaks were also observed in the bulk ChIP-seq data, indicating they are real regulations instead of artificial from scNanoCT\u0026amp;ATAC-seq (Extended Data Fig.\u0026nbsp;8e). In addition, the E2-H3K9me3 peaks mainly distributed close to the TSS, showing more cell-type specific characters (Extended Data Fig.\u0026nbsp;8f-g). Among these E2-H3K9me3 peaks, over 90% are overlapped with the cCREs and over 70% are intersected with active genome compartment (Extended Data Fig.\u0026nbsp;8h-i). Single-molecular analysis revealed highly interaction between the E2-H3K9me3 peaks and cis-regulatory elements (Extended Data Fig.\u0026nbsp;8j), suggesting this mode of H3K9me3 highly related to active gene regulation.\u003c/p\u003e \u003cp\u003eTo further identify the active regulatory characters of the E2-H3K9me3 modification, we compared the enriched TFs in GM12878 and K562 differently detected peaks (Extended Data Fig.\u0026nbsp;9a). In both cell types, the enriched motifs are highly consistent between ATAC and E2-H3K9me3 modules, resembling cell-type related TFs (Fig.\u0026nbsp;4e-f). For example, IRF1 as a master regulator in GM12878, showing top activity by ATAC module analysis, also ranked in the top list of E2-H3K9me3 enriched motifs. IRF1 also showed significant higher gene score, ATAC motif score and H3K9me3 motif score in GM12878 cells (Fig.\u0026nbsp;4e, Extended Data Fig.\u0026nbsp;9b-d). In contrast, GATA1 is detected as top regulator by both ATAC and E2-H3K9me3 in K562 cells. On the other hand, the classical E10-H3K9me3 modified regions are quite different to the regulatory elements, showing lower cell-type specificity (Extended Data Fig.\u0026nbsp;9b-d). All these results indicate that H3K9me3 not only constitutes heterochromatin, but also participates in active gene regulation.\u003c/p\u003e\n\u003ch3\u003eAccurate profiling of H3K9me3 modifications in transposon elements (TEs)\u003c/h3\u003e\n\u003cp\u003eTEs are evolutionary important elements, which accounts over half of the human genome\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. They have been reported to play as regulatory elements through epigenetic modifications, especially H3K9me3, in both developmental and disease contexts\u003csup\u003e\u003cspan additionalcitationids=\"CR30\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. However, previous NGS-based epigenetic profiling could not accurately assign the histone modification status of each TE copy on the genome, due to the high similarity of the sequences. Recently published scNanoSeq-CUT\u0026amp;Tag \u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e has claimed better coverage on full-length L1Hs profit from long read length of TGS. For all TE classes, there was no significant difference in the number of TE copies covered by classical NGS-based bulk ChIP-seq, scNanoSeq-CUT\u0026amp;Tag, and ScNanoCT\u0026amp;ATAC-seq (Fig.\u0026nbsp;4g, Extended Data Fig.\u0026nbsp;9e). However, for relative longer TE copies, the NGS-based bulk ChIP-seq detected lower H3K9me3 signals, indicating incomplete coverage. Specifically, for \u0026gt;\u0026thinsp;200bp SINEs, \u0026gt;\u0026thinsp;2000bp LINEs and \u0026gt;\u0026thinsp;1000bp LTRs, the two TGS based methods showed clearly higher H3K9me3 Scores (Fig.\u0026nbsp;4h). For example, we observed no read coverage on the long LINE copies within MS4A5 and downstream MS4A1 in the ENCODE ChIP-seq data. Thus, the TGS-based profiling exhibits obvious advantages in exploring the epigenetic status of these repetitive elements in the genome, even at single-cell resolution.\u003c/p\u003e \u003cp\u003eThen we checked the distribution of E2- and E10-H3K9me3 in the TE elements. As expected, in both GM12878 and K562 cells, the distribution ratios of E10-H3K9me3 peaks across the TE superfamilies were roughly the same, and the LINE accounted for the largest proportion\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e (Extended Data Fig.\u0026nbsp;9f). In comparison, the distribution of E2-H3K9me3 on different TE superfamilies switched dramatically from GM12878 to K562 cells. Further enrichment analysis also supported similar distribution of E10-H3K9me3 in different TE families, significantly enriched in L1, ERV1, ERVK and SVA (Fig.\u0026nbsp;4i). For the E2-H3K9me3, they highly enriched in MIR in GM12878 cells, while in ERV1, ERVK and SVA in K562 cells. This suggests that the regulation of active H3K9me3 modifications may also contribute to tumor development.\u003c/p\u003e\n\u003ch3\u003eEvaluating E-P interaction at single-molecular level\u003c/h3\u003e\n\u003cp\u003eIn similar to the previous TGS-based single-cell epigenome sequencing methods\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, scNanoCT\u0026amp;ATAC-seq also identifies the co-presence of two regulatory elements as they were captured on the same reads. As our integrated single-cell multi-epigenetic modalities accurately defined chromatin states in each cell population (Fig.\u0026nbsp;3a), we further focused on enhancers and promoters (see methods, Extended Data Fig.\u0026nbsp;10a), the most fundamental regulatory elements for gene transcription. We directly captured the interacting E-P pairs being supported by inter-reads spanning the corresponding peaks, and calculated the interaction score of each interacting enhancer by normalizing the supported reads to the total reads supporting the promoter (Fig.\u0026nbsp;5a). For both GM12878 and K562, we detected over ten thousand promoters and over thirty thousand enhancers, with the promoters largely shared (45.3%) compared to the enhancers (17.8%), indicating the enhancers showing higher degree of cell type specificity (Fig.\u0026nbsp;5b). GO analysis of the cell type specific promoters revealed lymphocyte differentiation pathways for GM12878 and myeloid cell differentiation pathways for K562, highly match the identities of the two types of cells (Fig.\u0026nbsp;5c). On the contrary, the common promoters enriched in basic pathways such as RNA metabolic and histone modification (Extended Data Fig.\u0026nbsp;10b), suggesting high accuracy of our element definition.\u003c/p\u003e \u003cp\u003eIn both cell types, the E-P pairs showed distance less than 10kb, while the none interactive enhancers often displayed longer distance to the nearest promoters (Extended Data Fig.\u0026nbsp;10c). Therefore the length of the scNanoCT\u0026amp;ATAC-seq library fragments limits the detection of long-distance E-P interactions. Within the limited distance (mostly less than 15kb), 68.29% promoters were detected with interacting enhancers (Extended Data Fig.\u0026nbsp;10d). The most interactive enhancers ranged around 2.5kb from the promoters (Extended Data Fig.\u0026nbsp;10e). Both the number of interacting enhancers and the global interacting frequencies regulated the transcriptional activities (Extended Data Fig.\u0026nbsp;10f). Further comparison of the E-P pairs between different cell types revealed that the vast majority enhancers regulating cell type specific promoters were also cell type specific (Fig.\u0026nbsp;5d, Extended Data Fig.\u0026nbsp;10g). Interestingly, even for the common promoters, they largely had different interacting enhancers in different cell types (Fig.\u0026nbsp;5d). Although cell type specific promoters showed the largest difference in E-P interaction frequency, the common E-P pairs had the highest interaction frequency in both cell types, suggesting that the conserved transcriptional regulation is the strongest (Fig.\u0026nbsp;5e). Theoretically, cell type-specific enhancers regulating the same promoters can serve as new therapeutic targets for disease treatment, especially for tumor treatment. For example, the lysine-specific demethylase KDM8 was commonly highly expressed in both K562 and GM12878 cells, while the upstream regulatory enhancers were completely different in the two types of cells (Fig.\u0026nbsp;5f, E20546 in K562 and E20547 in GM12878). Therefore, targeting E20546 should be specific to modulate KDM8 expression in K562 cells, without effect on GM12878 cells.\u003c/p\u003e \u003cp\u003eAs scNanoCT\u0026amp;ATAC-seq profiles the SVs as well as E-P interactions in the same cells, we figured out how genomic variations contribute to abnormal regulatory interactions in cancer. FAM78A, as a tumor associated gene highly expressed in hematologic disorders in the Human Protein Atlas, fused with chr13 in K562 cells (Extended Data Fig.\u0026nbsp;5c-d). Our data also captured E-P interactions between the chimera region (Fig.\u0026nbsp;5g). To prove the new enhancer promote FAM78A expression in cancer cells, we deleted the enhancer region using CRISPR system and measured the expression of FAM78A and its upstream gene on chr9 (Fig.\u0026nbsp;5h). The results validated the new E-P interaction formed by genomic translocation, demonstrating that scNanoCT\u0026amp;ATAC-seq can effectively capture the coordinated changes in genomic variations and regulatory elements in single cells.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eIdentifying genetic and epigenetic changes of tumor cells in LUAD\u003c/h2\u003e \u003cp\u003eWe applied scNanoCT\u0026amp;ATAC-seq to a LUAD sample to identify how genomic alternations cooperate with chromatin accessibility and H3K9me3 in tumorigenesis (Fig.\u0026nbsp;6a). A total of 1,920 cells were sequenced, and after quality control, we obtained 1,556 cells with matched ATAC and H3K9me3 data (see methods, Extended Data Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Using the ATAC module, we clearly identified 6 groups of cells: one cluster of normal epithelial cell, two clusters of tumor cells, one cluster of endothelial cells, one cluster of fibroblasts and one cluster of immune cells (Fig.\u0026nbsp;6b-c, Extended Data Fig.\u0026nbsp;11a-b). The tumor cells revealed the largest difference to the other cell types, while the non-epithelial cells were relatively mixed at H3K9me3 level, suggesting dramatic changes during tumorigenesis.\u003c/p\u003e \u003cp\u003eThen we analyzed the CNVs of each single cell, confirming copy number abnormalities in tumor cells (Fig.\u0026nbsp;6d). Meanwhile, the tumor cells were separated into two subclones based on the CNV pattern, which exactly corresponded to the two tumor clusters based on ATAC module. The tumor_1 cells gained more copies in the chromosome 5q32-q35 (Fig.\u0026nbsp;6d). The chromosome 8q region is a common variant area in LUAD\u003csup\u003e\u003cspan additionalcitationids=\"CR34\" citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Here, significant copy number increases of q22-q24 were also detected in both subclones of tumor cells. At single-cell level, we observed distinct CNVs from the two subclone of cells in this region (Fig.\u0026nbsp;6d). Moreover, we found the ATAC signals were highly related to the CNVs. For example,the oncogene MYC, which increased the copy number in both clones, showed higher ATAC signal in both clones (Fig.\u0026nbsp;6d-e, Extended Data Fig.\u0026nbsp;11c). CHRAC1, which significantly related to the survival of LUAD patients, gained genomic copies in tumor_1 cells, and only these tumor cells showed increased chromatin accessibility. On the other hand, the LUAD oncogene EIF3H, gaining genomic copies in tumor _2 cells, exhibited increased ATAC signals in these cells (Fig.\u0026nbsp;6d-e, Extended Data Fig.\u0026nbsp;11c). These results indicate that the epigenetic heterogeneity of tumor cells is highly consistent with the heterogeneity of genomic variations.\u003c/p\u003e \u003cp\u003eConsidering that the genomic copy number dosage may affect the calculation of the epigenetic signal value\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e, we discussed the chromatin accessibility changes by comparing to the genomic copy number changes (Fig.\u0026nbsp;6f). Take Tumor_1 cells as example, 277(7.89%) ATAC peaks showing comparable fold changes (FCs) in ATAC signal than copy number FCs, indicating these epigenetic changes are just dosage effect from genomic variations (Fig.\u0026nbsp;6g, Extended data Table S3). 2,708(77.13%) and 526(14.98%) peaks had higher and lower FCs in ATAC signal than copy numbers respectively, revealing de novo remodeling of these elements. The genes associated with the active increase of chromatin accessibility in tumor cells were mainly related to ERK1 and ERK2 cascade, Ras protein signal transduction and regulation of GTPase activity, supporting enhanced proliferation and metastasis facilitating tumor development. On the contrary, some RNA regulatory genes such as POLR2K, RNF19A and PABPC1, exhibited a decrease in chromatin accessibility-dependent dosage compensation in the copy number gaining regions. Thus we can clearly distinguish the coordinated regulation of the genome and epigenetic modifications in tumor cells using scNanoCT\u0026amp;ATAC-seq.\u003c/p\u003e \u003cp\u003eThe two tumor subclones did not show significant differences in H3K9me3 (Fig.\u0026nbsp;6b), so we merged these cells to analyzed how H3K9me3 changes in tumorigenesis. The H3K9me3 peaks in either tumor or normal cells were defined as E2-H3K9me3 and E10-H3K9me3 according to the accessibility (Extended Fig.\u0026nbsp;11d-e). For the changes of each type of H3K9me3 modification from normal to tumor cells, we analyzed how they related to the TEs (Extended Data Fig.\u0026nbsp;11f-g). The remodeling heterochromatin largely associated with LINE1 and ERVK elements, where most copies lost E10-H3K9me3 modification in tumor cells. This observation is consistent with the previous findings that tumorigenesis is accompanied by re-activation of heterochromatin\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Meanwhile, significant amount of LINE1 and LTR elements also gained H3K9me3 modification in tumor cells, implying highly precision of TE regulation (Extended Data Fig.\u0026nbsp;11g). Together, these results indicate the important role of H3K9me3 in LUAD tumorigenesis, and regulation of this modification is highly concentrated in TE elements such as LNE1 and ERVK.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eHistone modifications and chromatin structure provide regulatory diversity that is crucial during differentiation and homeostasis. Leveraging long-read sequencing, scNanoCT\u0026amp;ATAC-seq fundamentally advances single-cell multi-omics by enabling the simultaneous co-profiling of structural genomic variations (CNVs/SVs), multiple histone modifications (including traditionally challenging heterochromatic marks like H3K9me3), and chromatin accessibility within individual cells. This represents an opportunity to not only identify regions that exhibit cell-type-specific accessibility but also highlight elements whose acquisition of activating, repressive or heterochromatic signatures varies within a heterogeneous population. Our method first integrates data from multiple scNanoCT\u0026amp;ATAC-seq experiments together into a common manifold, generating coassay profiles for four histone modifications together with chromatin accessibility within individual cells. This allowed us to systematically define the chromatin states using ChromHMM\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, and comprehensively assess whether each genomic element is activated or repressed in single cells.\u003c/p\u003e \u003cp\u003eOne major challenge in human genetics is the prediction of the functional consequences of SVs on gene function and cis-regulatory networks\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. A key strength of scNanoCT\u0026amp;ATAC-seq lies in the concurrent acquisition of genetic and epigenetic data in single cells. This allows direct correlation of somatic mutations, e.g., SVs and indels, with local chromatin states, providing further insights into chromatin dynamics under specific genomic context. Specifically, we extract genomic mutations and identify the transitions in chromatin states between different cell types, thereby determining in which target cells the regulation of genetic variations occurs for diseases, and what functional elements of the target cells are affected.\u003c/p\u003e \u003cp\u003eSimilarly, we identified which cis-elements regulate a specific gene in a certain cell population through single-molecule E-P interactions. By comparing E-P interactions in different cells, we found that genes expressed in different cells have enhancer-shift situations, which might be caused by changes in the TF network under different cell states. This method advantage may help in the future development of specific targets for diseases such as tumors, for example, selecting conserved genes for cell survival but unique enhancers used in tumor cells as targets, which may specifically kill tumor cells while reducing side effects. However, the current single-molecule E-P interaction analysis also has certain limitations. Since the fragments recovered from the current library are basically within 10kb, it is difficult to capture E-P interactions over long distances. Fully resolving such architecture may require integration with complementary methods like Hi-C or long-read chromatin conformation capture.\u003c/p\u003e \u003cp\u003eOur data also reveal unexpected biological complexity, particularly regarding H3K9me3. We identified two functionally distinct modes: a classical repressive form (E10) in inaccessible, gene-poor heterochromatin, and a novel, active-associated form (E2) co-localizing with H3K4me3, H3K27ac, and accessible chromatin at promoters. These E2-H3K9me3 peaks bind cell-type-specific TFs (e.g., IRF1 in GM12878, GATA1 in K562) and show strong enrichment at cis-regulatory elements, suggesting a potential role in fine-tuning active transcription. Both types of H3K9me3 show dramatic changes in tumorigenesis, and these regulations highly related to TE elements. The mechanistic basis for this paradoxical H3K9me3 function and their role in developmental and disease context demands focused future investigation.\u003c/p\u003e \u003cp\u003ePrevious studies investigated the epigenetic regulations in cancer cells based on just one modality, such as ATAC-seq, ChIP-seq, etc. Theoretically, the calculation of these epigenetic signals is performed in euploidy. Tumor cells usually acquire large numbers of chromosomal CNVs, and such variations are overlooked in the analysis of epigenomics. By co-profiling of the genome and epigenome in single cells using scNanoCT\u0026amp;ATAC-seq, we show that many tumor-related epigenetic changes result from the dosage effect of genomic copy number. Besides, we propose gene activity maintaining under epigenetic-dependent dosage compensation, where the chromatin becomes less open with the increase in copy number. The epigenetic changes obtained by eliminating the influence of dosage may be more meaningful for therapeutic stratification of patients.\u003c/p\u003e \u003cp\u003eAlthough only one histone mark could be detected together with ATAC in the current version, several improvements can increase the histone modification modalities, i.e., using nanobody-based secondary antibody fused with pA/G-Tn5\u003csup\u003e18,19\u003c/sup\u003e, or pre-conjugating antibodies against histone marks with uniquely barcoded pA/G-Tn5 complexes\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. Besides, employing engineered reader domain-Tn5 fusions (e.g., HP1a chromodomain for H3K9me3)\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e can also bypass antibody limitations and improve the sensitivity.\u003c/p\u003e \u003cp\u003eIn summary, scNanoCT\u0026amp;ATAC-seq provides an unprecedented lens to examine the interplay between the genome and multi-layered epigenome in single cells. Its ability to concurrently map mutations, histone modifications, and chromatin accessibility opens new avenues for dissecting cellular heterogeneity in development and disease. Future refinements in scalability, modality diversity, and integration with spatial or 3D genomic data will further cement its role as a transformative tool in precision biology.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eEthics statement\u003c/h2\u003e \u003cp\u003eThis study was approved by the Ethics Committee of the First Affiliated Hospital of Guangzhou Medical University (Approval No: ES-2025-034-01). Sample tissues were collected from a LUAD patient with appropriate informed consent.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eCell culture and collection\u003c/h2\u003e \u003cp\u003eK562 and GM12878 cells were cultured in RPMI 1640 medium (Gibco, C11875500CP) supplemented with 10% fetal bovine serum (Procell, 164210-50) 2 mM L-glutamine (Gibco, 25030081) and 1% penicillin/streptomycin (Gibco, 15140122) at 37℃ in a humidified 5% CO₂ atmosphere. Both suspension cell lines can harvest directly by centrifugation at 200g for 5min, while GM12878 cells required additional pipetting to dissociate due to their naturally tendency to form multicellular aggregates.\u003c/p\u003e \u003cp\u003eThe K562-Cas9-BFP transgenic cell line was obtained from Dr. Nian Liu's laboratory (Tsinghua University). Cells were cultured in the normal K562 medium supplemented with 1% Non-Essential Amino Acids (NEAA; Gibco, 11140-050).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eSingle-cell suspension preparation of the LUAD tissue\u003c/h2\u003e \u003cp\u003eSurgically resected primary tumor and paired adjacent lung tissues of a LUAD patient were immediately preserved in pre-chilled MACS Tissue Storage Solution (Miltenyi Biotec, DS_130-100-008) and transported to the laboratory on ice. Tissues were minced into small pieces, followed by enzymatic dissociation in 5 ml DMEM/F12 medium (Gibco, #11330500BT) containing 2mg/mL collagenase D (Worthington, #LS004196) and 20U/mL DNase I (Worthington, #LS002004) at 37\u0026deg;C for 30\u0026ndash;45 min with gentle shaking to ensure uniform dispersion every 5\u0026ndash;10 min. After collagenase digestion, the tissue was thoroughly triturated using 1 ml pipette tip and allowed to settle for 2 min at room temperature. The supernatant containing liberated cells was transferred to a new tube and kept on ice. The remaining undigested tissue clumps were subjected to a second round of enzymatic digestion with 2ml 0.25% trypsin (Invitrogen, 25200056) at 37\u0026deg;C for another 15\u0026ndash;30 min. Inactivated trypsin by adding equal volume of 10% FBS plus DMEM/F12 medium and gently pipette to dissociate cells. Passed all the cell suspension through a 40 \u0026micro;m cell strainer. After red blood cell lysis, the cell suspension was subjected to negative selection using CD45 MicroBeads (Miltenyi Biotec, #130-045-801) to enrich non-immune cells according to the manufacturer's protocol. After depletion of immune cells, both primary and adjacent derived cells were used for scNanoCT\u0026amp;ATAC-seq library construction with antibody targeting H3K9me3.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eAntibodies\u003c/h2\u003e \u003cp\u003eThe following antibodies were used in our experimental procedures: rabbit anti-H3K4me3 (Millipore, 07-473, lot 3746382, 1:50 dilution), rabbit anti-H3K27ac (Abcam, ab177178, lot 1044138-34, 1:100 dilution), rabbit anti-H3K27me3 (Cell Signaling, 9733S, lot 16, 1:40 dilution) and rabbit anti-H3K9me3 (Abcam, ab8898, lot 1063772-1, 1:100 dilution).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eI5-pA/G_Tn5 and I7_Tn5 Loading\u003c/h2\u003e \u003cp\u003eThe single-stranded I5 adapter (TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG) and the Tn-ME (phos-CTGTCTCTTATACACATCT-NH3) were annealed at equimolar concentrations to form a 50 \u0026micro;M double-stranded I5-adapters. For assembly of the I5-pA/G_Tn5, 20 2 \u0026micro;L pA/G-Tn5 transposase (Vazyme, S604) and 3.5 \u0026micro;L I5-adapters were mixed with 14 \u0026micro;L coupling buffer by gently pipetting, and incubated at 30\u0026deg;C for 1 h. Similarly, the single-stranded I7 adapter (GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG) and the Tn-ME sequence were annealed at equimolar concentrations to form a 50 \u0026micro;M double-stranded I7 adapters. For assembly of the I5 _Tn5, 20 \u0026micro;L Tn5 transposase (Novoprotein, M045) and 4 \u0026micro;L I7-adapters were mixed directly by gently pipetting, and incubated at room temperature for 1 h. \u003cb\u003eScNanoCT\u0026amp;ATAC-seq experimental workflow\u003c/b\u003e\u003c/p\u003e \u003cp\u003e200,000-500,000 viable cells were collected. First, single-cell suspensions were incubated with primary antibodies against specific histone modification at 4\u0026deg;C overnight. Subsequently, these cells were incubated with prebound second antibody tethered I5-pA/G_Tn5 transposase to recognize and amplify the primary antibodies. After washed out excess antibody and I5_pA/G-Tn5 thoroughly, cell pellets were incubated by I7_Tn5 transposase targeting accessible chromatin sites and both I5_pA/G-Tn5 and I7_Tn5 were activated by Mg2\u0026thinsp;+\u0026thinsp;ions at 37\u0026deg;C for 1 h. Tagmentation was stopped by 20mM EDTA and stained with 10 \u0026micro;g/mL DAPI, the nuclei were immediately sorted using a flow cytometer. Single nucleus was deposited into each well of a 96-well plate pre-loaded with 2 \u0026micro;L of lysis buffer containing 50mM Tris-HCl (pH 8.0), 50 mM NaCl, 1.5 \u0026micro;M of both the I5 and I7 inner barcode primers. Nuclei were lysed to release DNA fragments at 50℃ for 60 min. SDS(contained in the lysis buffer which would inactivate the subsequent PCR polymerase) was quenched with 2 \u0026micro;L 10% Tween20. First round of PCR amplification was performed directly in 10 \u0026micro;L reaction buffer each well containing 0.25U Tks Gflex DNA Polymerase (TAKARA, R060A) as follows: 68℃ for 10min; 95\u0026deg;C for 1 min; 14 cycles of 94\u0026deg;C for 15s, 63\u0026deg;C for 30 s, 68\u0026deg;C for 8 min and finally 68\u0026deg;C for 5 min.\u003c/p\u003e \u003cp\u003ePCR products of each row with different inner barcodes were pooled and purified with 0.7X AMPure XP beads (Beckman Coulter; A63882) and eluted with 21 \u0026micro;L H2O. The second round of amplification was carried out in 50 \u0026micro;L reaction mixture with 0.25U Tks Gflex DNA Polymerase and 1 \u0026micro;M different Outer barcode Primers. The PCR program was as follows: 95\u0026deg;C for 1 min, then 5 cycles of 94\u0026deg;C for 15 s, 63\u0026deg;C for 30 s, 68\u0026deg;C for 8 min and finally 68\u0026deg;C for 5 min. After two rounds of PCR amplifications, DNA fragments derived from different cell in each well of 96-well plate were tagged with different inner and outer barcode combination, pooled and purified with 1X AMPure XP beads into a single library.\u003c/p\u003e \u003cp\u003eLibraries from 8\u0026ndash;10 plates with different barcode combination were pooled together, preceded to ONT library construction with sequencing kit (Oxford Nanopore Technologies, SQK-LSK114) following the manufactures\u0026rsquo; instruction and sequenced on one PromethION Flow Cell R10 Version (Oxford Nanopore Technologies, FLO-PRO114M).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eScNanoATAC-seq library preparation and sequencing\u003c/h2\u003e \u003cp\u003escNanoATAC-seq procedure was adapted from previous work by Hu \u003cem\u003eet al\u003c/em\u003e.\u003csup\u003e21\u003c/sup\u003e. Briefly, 200,000 viable cells were lysed with 200 \u0026micro;L Omni-ATAC lysis buffer as described by Corces et al.\u003csup\u003e40\u003c/sup\u003e on ice for 3 min and rinsed by adding 1 ml of ATAC-seq wash buffer. Immediately, nuclei were resuspended with 50 \u0026micro;L tagmentation mixture with 1 \u0026micro;M I7_Tn5 transposase and incubated at 37\u0026deg;C, 800 rpm for 30 min. Transposition was stopped by 20mM EDTA. Nuclei were stained with DAPI and sorted immediately into each well of a 96-well plate pre-loaded with 2 \u0026micro;L lysis buffer with only 1.5 \u0026micro;M I7 inner barcode primers. The subsequent treatment and library construction steps were identical to those described above for the scNanoCT\u0026amp;ATAC-seq.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eScNanoSeq-CUT\u0026amp;Tag library preparation and sequencing\u003c/h2\u003e \u003cp\u003e200,000-500,000 viable cells as single cell suspension were incubated with primary antibodies against specific histone modification at 4\u0026deg;C overnight. Subsequently, these cells were incubated with prebound second antibody tethered I5-pA/G_Tn5 transposase to recognize and amplify the primary antibodies. After washed out excess antibody and I5_pA/G-Tn5 thoroughly, cell pellets were incubated by 50 \u0026micro;L tagmentation buffer with 10mM Mg2\u0026thinsp;+\u0026thinsp;to activate transposition at 37\u0026deg;C, 500 rpm for 1 h. After the reaction was terminated by the addition of 20 mM EDTA, the nuclei were stained with DAPI and immediately sorted into a 96-well plate. Each well was pre-loaded with 2 \u0026micro;L of lysis buffer supplemented with only 1.5 \u0026micro;M I5 inner barcode primers. The subsequent treatment and library construction steps were identical to those described above for the scNanoCT\u0026amp;ATAC-seq.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eBulk NGS-based CUT\u0026amp;Tag-seq\u003c/h2\u003e \u003cp\u003eBriefly, 50, 000 cells were collected freshly and centrifuged at 300g for 5 min at room temperature. Cells pellets were incubated with activated ConA beads Pro and proceeded to the following antibody incubation and pA/G-Tnp transposition. Targeted DNA fragments were extracted for library preparation using the Hyperactive Universal CUT\u0026amp;Tag Assay Kit for Illumina Pro (Vazyme, TD904). All NGS libraries were sequenced using SURFSeq 5000 sequencer with PE150 reads.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eFunctional validation of the SV-derived FAM78A E-P interaction\u003c/h2\u003e \u003cp\u003eAfter the identification of chromosome fusion event of chr13:108M and chr9: FAM78A, we annotated active enhancers and promoters to fide out that enhancer E15695 may form novel E-P interactions with the promoter of FAM78A to regulatory the expression of this gene. We designed two pairs of single guide RNAs (sg13 and sg24) using the online tool (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://crispor.org\u003c/span\u003e\u003cspan address=\"http://crispor.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) targeting the flanking regions of E15695. Oligonucleotides were synthesized by Sangon Biotech (Shanghai, People's Republic of China). The tandem sgRNA cassette was introduced into the lentivirus vector (pLV3-Amp_hU6-esgRNA_EF1a-NeoR-T2A-EGFP-backbone) using multiplex PCR followed by homologous recombination. Lentivirus was produced in 293T cells, and the concentrated virus was then used to infect the K562-Cas9-BFP cells. GFP and BFP double positive cells were sorted by FACS and continued to be cultured for subsequent mRNA extraction and RT-qPCR detection for the quantification of FAM78A gene expression. RAPGEF1 gene located upstream of the chr9: FAM78A breakpoint was also quantified as a non-regulated control.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eReverse-transcription and real-time quantitative PCR\u003c/h2\u003e \u003cp\u003eCells were collected to extract RNA using RNeasy Mini Kit (Qiagen, 74104). mRNA was reverse-transcribed into cDNA using Maxima H Minus Reverse Transcriptase (ThermoFisher Scientific, EP0753). Real-time quantitative PCR was performed using ChamQ SYBR qPCR Master Mix (Vazyme, Q311-02) according to the manuals on Quantagene q225 real-time PCR system (Kubo Technology) in 10 \u0026micro;L reaction mixture with 3 replicates.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eData preprocessing\u003c/h2\u003e \u003cp\u003eFor scNanoCT\u0026amp;ATAC-seq data, Nanopore sequencing raw signals were converted to DNA sequences using the high-accuracy model \u0026ldquo;dna_r10.4.1_e8.2_400bps_hac\u0026rdquo; of Dorado (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/nanoporetech/dorado/\u003c/span\u003e\u003cspan address=\"https://github.com/nanoporetech/dorado/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) v0.2.0 software. Reads with quality scores less than 7 were discarded. To demultiplex single-cell files from raw sequences, we used the nanoplexer (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/hanyue36/nanoplexer\u003c/span\u003e\u003cspan address=\"https://github.com/hanyue36/nanoplexer\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) v0.1 software with default parameters. Demultiplexing was performed twice on the outer and inner cell barcodes. We used the cutadapt\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003ev4.1 software with a minimal sequence length of 12 bp and a maximum error rate of 0.2 to trim the adaptors sequentially from both ends of the reads and record the adaptor information. The adaptor sequences include \u0026lsquo;GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG\u0026rsquo; employed to tag transposase-accessible chromatin and \u0026lsquo;TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG\u0026rsquo; employed to tag chromatin modifications. After trimming the cell barcodes and adaptors, reads with read length less than 100 bp were discarded. Reads were aligned to the human reference genome (GRCh38) using minimap2\u003csup\u003e42\u003c/sup\u003e v2.26 with parameters \u0026ldquo;-ax map-ont --MD\u0026rdquo;. Alignments were subjected to sorting by coordinate, filtering by a minimal mapping quality (MAPQ) of 30, and PCR duplicate removal using SAMtools\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e v1.13.\u003c/p\u003e \u003cp\u003eFor in-house scNanoATAC-seq and scNanoSeq-CUT\u0026amp;Tag data, the procedures and parameters of basecalling, demultiplexing, adaptor trimming, mapping and quality control were identical to those described above.\u003c/p\u003e \u003cp\u003eFor bulk CUT\u0026amp;TAG-seq data, we used the Trimmomatic\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e v0.39 to trim reads and remove adaptor sequences of Illumina platforms with the parameters: PE MINLEN:20 CROP:100. Read mapping were performed using bowtie2\u003csup\u003e45\u003c/sup\u003e v2.4.2 with parameters \u0026ldquo;--mm -X2000\u0026rdquo;. Low-quality alignments and duplicated alignments were also removed using SAMtools.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eExtraction of chromatin accessibility and modification signals\u003c/h2\u003e \u003cp\u003eThe two ends of the alignments corresponded to open or modified chromatin sites. The aligned bam files were converted into bed format files using the \u0026lsquo;bamtobed\u0026rsquo; command of BEDtools\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e v2.30.0. The genome coordinates of two ends were extracted using the \u0026lsquo;flank\u0026rsquo; command of BEDtools with a 1 bp flanking size. By integrating the prior cell barcode and adapter information, we derived two chromatin mark signals for each cell, which were output to fragment files in bed format. For downstream analyses, this information was also added to the alignment files in the \u0026lsquo;CB:Z\u0026rsquo; tag. Cells with a total count of fragments below 5,000 were considered to have insufficient data and were filtered out.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eSingle-cell chromatin mark analysis using ArchR\u003c/h2\u003e \u003cp\u003eThe chromatin mark data we stratified into active marks (ATAC, H3K4me3, H3K27ac) and repressive marks (H3K9me3, H3K27me3) based on their impact on gene regulation. Fragment data of an individual mark was loaded and converted into Arrow files using the \u0026lsquo;createArrowFiles\u0026rsquo; function of ArchR\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e v1.0.1. Filtering low-quality single-cell data was performed with the following parameters: minTSS\u0026thinsp;=\u0026thinsp;1/0.1 (respectively for active/repressive marks), minFrags\u0026thinsp;=\u0026thinsp;1000, addTileMat\u0026thinsp;=\u0026thinsp;TRUE, addGeneScoreMat\u0026thinsp;=\u0026thinsp;TRUE. We additionally identified and discarded potential doublets by using the \u0026lsquo;add DoubletScores\u0026rsquo; function.\u003c/p\u003e \u003cp\u003eAll the Arrow files were load and consolidated into an ArchR project for analysis. Dimensionality reduction was performed by the \u0026lsquo;addIterativeLSI\u0026rsquo; function with the following parameters: useMatrix = \"TileMatrix\", name = \"IterativeLSI\", iterations\u0026thinsp;=\u0026thinsp;2, dimsToUse\u0026thinsp;=\u0026thinsp;1:30. Clustering was performed using the \u0026lsquo;addClusters\u0026rsquo; function with \u0026lsquo;method\u0026rsquo; set as \u0026lsquo;Seurat\u0026rsquo;. An UMAP embedding was obtained by the \u0026lsquo;addUMAP\u0026rsquo; function with default parameters.\u003c/p\u003e \u003cp\u003eTo annotate the cell identity of clusters, we firstly calculated gene score using the \u0026lsquo;addGeneScoreMatrix\u0026rsquo; function on an individual mark. Here, the gene score serves as a proxy for the strength of the chromatin mark around the gene region. We calculated the marker genes of individual cell clusters by the \u0026lsquo;getMarkerFeatures\u0026rsquo; function with the following parameters: useMatrix = \"GeneScoreMatrix\", groupBy = \"Clusters\", bias\u0026thinsp;=\u0026thinsp;c(\"TSSEnrichment\", \"log10(nFrags)\"), testMethod = \"wilcoxon\". Marker genes were defined as those with log2-transformed fold change (log2FC)\u0026thinsp;\u0026ge;\u0026thinsp;1 and p-value\u0026thinsp;\u0026le;\u0026thinsp;0.05. We annotated cell clusters by known cell type marker genes.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003ePeak calling\u003c/h2\u003e \u003cp\u003eTo reduce the effect of the sparsity of single-cell data on peak identification, fragment data were partitioned based on annotated cell types and further used for peak calling. Each fragment was extended by 75 bp on both sides to be biologically and computationally consistent with the short-read ATAC and CUT\u0026amp;TAG data. Extended fragments were used for peak calling. For ATAC data, we used the \u0026lsquo;callpeak\u0026rsquo; command of MACS2\u003csup\u003e48\u003c/sup\u003e v2.2.9.1 with parameters \u0026ldquo;-f BED -q 0.05 --keep-dup all -B --SPMR\u0026rdquo;. For histone modification data, we used SEACR\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e v1.3 with parameters \u0026ldquo;non stringent 0.05\u0026rdquo;. Additionally, peaks supported by more than 2.5% of cells were considered reliable and retained for subsequent analysis.\u003c/p\u003e \u003cp\u003eFor bulk CUT\u0026amp;TAG-seq data, the parameters of peak calling were identical to those described above.\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eDifferential peak analysis of chromatin marks\u003c/h2\u003e \u003cp\u003eFor parallel comparison, peaks from each cell type were consolidated into a unified list. Using the BEDtools \u0026lsquo;merge\u0026rsquo; command, overlapping peaks were merged into single representative peaks while non-overlapping peaks remained unchanged. We then annotated whether each peak was identified in one or multiple cell types. The peak list was loaded into ArchR using the \u0026lsquo;addPeakSet\u0026rsquo; function, and their signal strength in each cell was calculated by the \u0026lsquo;addPeakMatrix\u0026rsquo; function.\u003c/p\u003e \u003cp\u003eWe calculated the differentially expressed peaks between two cell clusters by the \u0026lsquo;getMarkerFeatures\u0026rsquo; function with the following parameters: useMatrix = \"PeakMatrix\", groupBy = \"Clusters\", bias\u0026thinsp;=\u0026thinsp;c(\"log10(nFrags)\"), testMethod = \"wilcoxon\", useGroups = \u0026ldquo;cell_type1\u0026rdquo;, bgdGroups = \u0026ldquo;cell_type2\u0026rdquo;. The highly expressed peaks in cell type 1 required log2FC\u0026thinsp;\u0026ge;\u0026thinsp;1 and p-value\u0026thinsp;\u0026le;\u0026thinsp;0.05, whereas those in cell type 2 required log2FC \u0026le; -1 and p-value\u0026thinsp;\u0026le;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eFunction enrichment and motif analysis of peaks\u003c/h2\u003e \u003cp\u003eThe functional GO analysis of the genes linked to marker or differential peaks was conducted by R package ClusterProfiler\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e v4.6.2 with a p-value cut-off of 0.01. Qualified GO terms with top gene ratios were visualized. In addition, similar GO terms were consolidated, with one representative term being kept. We extracted genomic regions spanning 1 kb upstream and downstream of TSS of genes. For active marks, we associated nearby peaks with corresponding genes using BEDtools \u0026lsquo;intersect\u0026rsquo; command.\u003c/p\u003e \u003cp\u003eThe motif annotation of peaks was performed using the \u0026lsquo;addMotifAnnotations\u0026rsquo; function with cisbp database. On the basis of the results of differential peak analysis, motif enrichment was performed using the \u0026lsquo;peakAnnoEnrichment\u0026rsquo; function. To extract motifs enriched in cell type 1, the parameters were set as follows: peakAnnotation = \"Motif\", cutOff = \"Pval\u0026thinsp;\u0026lt;\u0026thinsp;=\u0026thinsp;0.05 \u0026amp; Log2FC\u0026thinsp;\u0026gt;\u0026thinsp;=\u0026thinsp;0.5\". To extract motifs enriched in cell type 2, the parameters were set as follows: peakAnnotation = \"Motif\", cutOff = \"Pval\u0026thinsp;\u0026lt;\u0026thinsp;=\u0026thinsp;0.05 \u0026amp; Log2FC \u0026lt;= -0.5\". Using the \u0026lsquo;getFeatures\u0026rsquo; and \u0026lsquo;plotGroups\u0026rsquo; function, the motif score of cells were extracted and visualized across cell types using boxplots. For visualization, we extracted the position weight matrix (PWM) of each motif using the \u0026lsquo;getPeakAnnotation\u0026rsquo; in ArchR, which was further loaded into the R package \u0026lsquo;ggseqlogo\u0026rsquo;\u003csup\u003e51\u003c/sup\u003e v0.2 to generate sequence logos. As an alternative approach, bed files containing peaks can also be used as input to the HOMER (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://homer.ucsd.edu/homer/\u003c/span\u003e\u003cspan address=\"http://homer.ucsd.edu/homer/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) software v5.1 for enriched motif calculation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003eEvaluation of the throughput of multi-modal data at single-cell level\u003c/h2\u003e \u003cp\u003eThe four multi-modal data types comprised ATAC-seq (I7 adaptor) in combination with the following modifications: H3K4me3, H3K27ac, H3K9me3, and H3K27me3 (I5 adaptor). Those read ends without matched adaptors were named unclassified (UN).\u003c/p\u003e \u003cp\u003eFirstly, we quantified the count, length, and total data size of sequencing reads per cell. After adaptor trimming, we calculated the count and proportion of the I5, I7 and UN fragments. Based on the adaptors, reads were classified into 5 types: I5_I5, I5_I7, I7_I7, UN_I5 and UN_I7. We quantified the count, proportion and length of modality-tagged reads. After read mapping, the count of uniquely mapped reads and fragments was quantified per cell. Then, the mapping ratio was defined as the ratio of aligned reads (MAPQ\u0026thinsp;\u0026gt;\u0026thinsp;30) to the total number of modality-tagged reads. The total number of mapped bases per cell were calculated by the \u0026lsquo;depth\u0026rsquo; command of SAMtools. The mapping coverage was defined as the number of mapped bases divided by the total size of the reference genome. For comparative analysis, we computed the statistics for in-house scNanoATAC-seq and scNanoSeq-CUT\u0026amp;Tag data in a manner consistent with the aforementioned methods.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003eEvaluation of efficiency, reliability and reproducibility of scNanoCT\u0026amp;ATAC-seq\u003c/h2\u003e \u003cp\u003eWe first evaluated individual modalities using the fractions of read ends in peaks (FRIP) and the transcriptional start site (TSS) enrichments. For FRIP, the peaks called from all cells of the modality served as the target regions. Using the \u0026lsquo;intersect\u0026rsquo; command of BEDtools, we calculated the number of fragments falling within these peaks, which was then divided by the total number of fragments to obtain the FRIP. The TSS enrichments of each cell had already been calculated in previous quality control step of ArchR. The TSS distribution profile were extracted using the \u0026lsquo;plotTSSEnrichment\u0026rsquo; function of ArchR.\u003c/p\u003e \u003cp\u003eSubsequently, we evaluated the technical reproducibility across replicates. For each cell line and each set of multimodal data, we generated a comparable number of single-cell datasets in two separate experimental batches (196 cells per batch). We partitioned the genome into consecutive 10-kb bins. For each bin, we counted the number of tagged-fragments from each of the two batches. The data was structured into a matrix where rows represented bins, columns represented the two batches, and values contained the fragment counts. We then computed the correlation of fragment counts across all bins between the two batches using Pearson correlation method. At the peak level, the overlapped peaks between two batches were identified by BEDTools \u0026lsquo;intersect\u0026rsquo; command, and the ratio of overlapped peaks to all peaks from individual batch were calculated as an indicator of reproducibility.\u003c/p\u003e \u003cp\u003eFor cross-platform validation, we used the peaks called by deep-sequencing bulk NGS-based data as a benchmark for evaluating the peak precision and recall. Using the \u0026lsquo;intersect\u0026rsquo; command of BEDtools, we identified the scNanoCT\u0026amp;ATAC-seq peaks overlapped with NGS-based peaks as true-positive (TP) peaks. Meanwhile, the unique peaks of scNanoCT\u0026amp;ATAC-seq were defined as false-positive (FP) peaks, and the unique peaks of NGS data were defined as false-negative (FN) peaks. Then, we calculated the precision of peaks as TP/(TP\u0026thinsp;+\u0026thinsp;FP) and the recall of peaks as TP/(TP\u0026thinsp;+\u0026thinsp;FN). In addition to the global assessment, we further examined the precision and recall ratio of peaks by leveraging datasets of increasing cell numbers. Similarly, the correlation between bins and the consistence between peaks from the two techniques were evaluated as described previously.\u003c/p\u003e \u003cp\u003eTo visualize chromatin marks across the genome, we used deepTools\u003csup\u003e\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e v3.5.5 to compute and plot signal distribution heatmaps spanning 1 kb/5 kb regions centered on peak summits. Genomic coverage tracks in bigwig format were generated with the \u0026lsquo;getGroupBW\u0026rsquo; function in ArchR with parameters \u0026ldquo;tileSize\u0026thinsp;=\u0026thinsp;100, maxCells\u0026thinsp;=\u0026thinsp;10000\u0026rdquo;. For NGS-based data, we generated the BedGraph files using the BEDtools \u0026lsquo;genomecov\u0026rsquo; command, then converted them to bigwig format using bedGraphToBigWig\u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e v4. The bigwig files were loaded into Integrative Genomics Viewer (IGV) for visualization.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003eGenomic characteristics and distribution of multimodal data\u003c/h2\u003e \u003cp\u003eIt is widely acknowledged that the five modalities display characteristic genomic distributions, which also serves as an evaluation metric. Firstly, we evaluated the relative enrichment of individual modalities to NGS-based peaks as benchmark. In brief, we quantified the count of modality-tagged fragments overlapping the peaks. The counts were normalized by their corresponding library sizes and log2-transformed, generating a matrix where rows represent peaks and columns contain the relative enrichment scores for the five modalities.\u003c/p\u003e \u003cp\u003eWe then annotated the genomic location of peaks using the R package ChIPseeker\u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e v1.34.1. We also annotated the candidate cis-regulatory elements (cCREs)\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e that covered by peaks using the BEDtools \u0026lsquo;intersect\u0026rsquo; command.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStructural variation (SV) detection and evaluation\u003c/h3\u003e\n\u003cp\u003eWe performed SV calling on single-cell bam files with cuteSV\u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e v1.0.10. The parameters were set as follow: --max_cluster_bias_INS 100 --diff_ratio_merging_INS 0.3 --max_cluster_bias_DEL 100 --diff_ratio_merging_DEL 0.3 --min_support 1. For each cell, any SV supported by at least one aligned read was retained and recorded in the vcf file. The five SV types were: deletions (DEL), insertions (INS), duplications (DUP), inversions (INV), and breakends (BND). For the first four SV types, we extracted genomic coordinates from vcf files and converted them into bed format. Within each cell type, all SVs were merged based on their type and genomic coordinates. We then quantified the number of supporting cells for each merged SV. The supporting cell count served as a reliability metric for downstream filtering. BND variants represent genomic rearrangements involving two distant genomic loci, with each VCF record containing paired coordinates. For integration, we binned these coordinates at 1000 bp resolution. BND events sharing identical binned coordinates were then consolidated. Subsequent quantification of supporting cells and filtering were performed consistent with the aforementioned methods.\u003c/p\u003e \u003cp\u003eFor benchmarking, we downloaded K562 bulk whole-genome data (WGS) generated by Nanopore sequencing from public sources\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. Data were processed through the same mapping and SV calling pipeline, and only the SVs supported by more than 5 reads were retained to generate a high-confidence benchmark SV set. We identified the SVs of scNanoCT\u0026amp;ATAC-seq overlapped with SVs of WGS as true-positive (TP). Meanwhile, the unique SVs of scNanoCT\u0026amp;ATAC-seq were defined as false-positive (FP), and the unique SVs of WGS were defined as false-negative (FN). Then, we calculated the precision of SVs as TP/(TP\u0026thinsp;+\u0026thinsp;FP) and the recall of SVs as TP/(TP\u0026thinsp;+\u0026thinsp;FN). In addition to the global assessment, we further examined the precision and recall ratio of SVs supported by increasing cell numbers.\u003c/p\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003eCNV detection and evaluation\u003c/h2\u003e \u003cp\u003eWe performed CNV quantification on single-cell bam files with Control-FREEC\u003csup\u003e\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e v11.6b. The parameters were set as follow: ploidy\u0026thinsp;=\u0026thinsp;2, breakPointThreshold\u0026thinsp;=\u0026thinsp;.8, window\u0026thinsp;=\u0026thinsp;1000000, minExpectedGC\u0026thinsp;=\u0026thinsp;0.30, maxExpectedGC\u0026thinsp;=\u0026thinsp;0.60, sex\u0026thinsp;=\u0026thinsp;XY, inputFormat\u0026thinsp;=\u0026thinsp;BAM, mateOrientation\u0026thinsp;=\u0026thinsp;0. The WGS data were processed through the same pipeline. We calculated the average CNV ratio of each 1 Mb window across all cells as a global metric, which was compared to the CNV ratio from WGS. The CNV ratio for each genomic window in every single cell was visualized as a heatmap, annotated with corresponding cell types and library sources.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003eAssigning chromatin states from multimodal data at the bulk level\u003c/h2\u003e \u003cp\u003eTo integrate information from multiple modalities, we used the multivariate HMM introduced in ChromHMM\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e v1.26. All the modality-tagged fragments from individual cells were merged to produce pseudo-bulk bed files. The configuration file documenting the cell type, modality, and file paths was processed by the \u0026lsquo;BinarizeBed\u0026rsquo; command with default parameters. The resulting files binarized at 200 bp resolution were loaded into the \u0026lsquo;LearnModel\u0026rsquo; function to model the combinatorial states and spatial patterns from five chromatin marks. The number of possible states were set as 10. Ten states were functionally annotated based on the high probability chromatin marks and the enrichment in specific genomic regions in state.\u003c/p\u003e \u003cp\u003eWe obtained the RNA-seq data of cell lines from ENCODE\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e to validate the biological relevance of ChromHMM-derived chromatin states. We extracted all genes within the 10 chromatin states, followed by a parallel comparison of expression levels across these states. As hypothesized, genes in active states exhibited significantly higher expression than those in repressive states.\u003c/p\u003e \u003cdiv id=\"Sec33\" class=\"Section3\"\u003e \u003ch2\u003eAssigning chromatin states from multimodal data at the single-cell level\u003c/h2\u003e \u003cp\u003eSingle-cell chromatin state analysis was performed using scChromHMM, an extention of bulk ChromHMM framework established in prior work\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, with specific adaptations for the scNanoCT\u0026amp;ATAC-seq dataset. The most critical step, termed \u0026ldquo;Anchor\u0026rdquo;, involves leveraging a reliable single-cell annotation set to associate multiple modalities of data with individual cells. Using the ATAC layer as a reliable reference, in this study, the modalities co-assayed in the same cell are intrinsically linked. For other modalities, we used the \u0026ldquo;FindTransferAnchors\u0026rdquo; algorithm in Seurat\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e v4 to identify anchor correspondences. Importantly, the number of modalities associated with individual cells is variable, with some cells encompassing two or more data types.\u003c/p\u003e \u003cp\u003eScChromHMM requires four input files. The pseudo-bulk HMM model parameters generated previously were used. Since scChromHMM is currently limited to processing 10x Genomics cell barcodes, we generated compatible pseudo-barcodes for each scNanoCT\u0026amp;ATAC-seq cell library. The fragment files, reference cell barcodes and anchors list were then generated according to the manual. We also maintained a cross-reference table between original and converted barcodes to ensure data provenance throughout the analysis.\u003c/p\u003e \u003cp\u003eWe sequentially executed the \u0026lsquo;hmm\u0026rsquo; and \u0026lsquo;transform\u0026rsquo; modules using default parameters. The output of scChromHMM is the posterior probabilities of each chromatin state for every reference cell at 200 bp resolution. In this study, scChromHMM analysis identified two primary chromatin states: Active and Repressed, serving as quantification metrics for comprehensive assessment of chromatin activity.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec34\" class=\"Section3\"\u003e \u003ch2\u003eEvaluation of scChromHMM application on scNanoCT\u0026amp;ATAC-seq\u003c/h2\u003e \u003cp\u003eFor evaluation, we calculated the average active/repressed probabilities across all cells, counted and normalized the number of modality-tagged fragments of each 1 kb bin. The Pearson correlation between scChromHMM probabilities and individual modality signals were calculated. Then, we defined the A/R score as the ratio of active to repressed probabilities, and the regions with A/R score\u0026thinsp;\u0026ge;\u0026thinsp;2 were classified as active regions, while those with A/R score\u0026thinsp;\u0026le;\u0026thinsp;0.5 were designated as repressed regions. We calculated the mean A/R scores by scChromHMM across 10 chromatin states of ChromHMM, demonstrating their concordance.\u003c/p\u003e \u003cp\u003eTo assess the potential of the A/R score as predictor for gene activity, we stratified genes based on their TPM values in ENCODE RNA-seq datasets into four categories: no expression (N/E, TPM\u0026thinsp;=\u0026thinsp;0), low-level (Low, \u0026le; 25th percentile), medium-level (Mid, \u0026gt; 25th and \u0026le;\u0026thinsp;90th percentile), and high-level (High, \u0026gt; 90th percentile). We visualized the distribution of A/R scores across gene groups with boxplots and performed one-way ANOVA with Tukey's test for multiple comparisons.\u003c/p\u003e \u003cp\u003eWe next performed cell clustering based on the A/R score, an analogous process as single-modality analysis. For the \u0026plusmn;\u0026thinsp;1 kb region surrounding TSS of each gene in each cell, we computed the A/R scores and integrated into a gene activity/expression matrix. The matrix were loaded and converted into a Seurat object using the \u0026lsquo;CreateSeuratObject\u0026rsquo; function of Seurat with the parameters for quality control: min.cells\u0026thinsp;=\u0026thinsp;5, min.features\u0026thinsp;=\u0026thinsp;300. After choosing the most variable features and data scaling, we performed the Principal Component Analysis (PCA) for dimensionality reduction, cell clustering, and UMAP embedding with default parameters. We evaluated the concordance in cell clustering between the A/R score-based and ATAC-based approaches, and visualized the A/R score of cell line marker genes at the single-cell resolution.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eH3K9me3 enrichment on TEs\u003c/h3\u003e\n\u003cp\u003eRepetitive regions annotated by RepeatMasker were downloaded from UCSC\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e, from which we extracted five major classes of TEs (LINE, LTR, Retroposon, SINE, DNA) and their respective subcategories. We annotated the TEs that covered by peaks using the BEDtools \u0026lsquo;intersect\u0026rsquo; command. We compared the enrichment of TEs in H3K9me3-modified versus unmodified regions through applying chi-square test to assess statistical significance and calculating the odds ratios. We also quantified the types, counts, and proportions of TEs overlapping with H3K9me3-marked regions.\u003c/p\u003e \u003cp\u003eTo evaluate the advantages of long-read sequencing in H3K9me3 profiling in TEs/repeats, we collected the bigwig files of three technologies: NGS-based ChIP-seq\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e, scNanoSeq-CUT\u0026amp;Tag, and scNanoCT\u0026amp;ATAC-seq.\u0026nbsp;Using the \u0026lsquo;bigWigAverageOverBed\u0026rsquo; command of UCSC\u003csup\u003e\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e, we extracted the signal values for each TE and normalized them for parallel comparison. We counted the number of TEs uniquely and commonly identified across platforms, and assessed their strengths in TEs of increasing lengths.\u003c/p\u003e\n\u003ch3\u003eCategorization and characterization of H3K9me3 peaks\u003c/h3\u003e\n\u003cp\u003eIn the chromatin state analysis, we identified not only the canonical H3K9me3 signals localized to heterochromatin (state E10) but also a novel category of H3K9me3 signals occurring at active transcription start sites (state E2). We characterized these two categories with respect to their peak length, correlation with gene expression, motif enrichment, TEs enrichment, and cCREs enrichment, following the same methods as described previously. We also downloaded the A/B compartment files from Rao et al.\u003csup\u003e59\u003c/sup\u003e, which is processed based on previous version of human reference genome (GRCh37). Therefore, we lifted the coordinates of H3K9me3 peaks to GRCh37 using \u0026lsquo;liftOver\u0026rsquo; (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://genome.ucsc.edu/cgi-bin/hgLiftOver\u003c/span\u003e\u003cspan address=\"http://genome.ucsc.edu/cgi-bin/hgLiftOver\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and calculated their distribution on A/B compartments using the \u0026lsquo;intersect\u0026rsquo; command of BEDtools.\u003c/p\u003e \u003cp\u003eWe further measured the co-localization between E2/E10 H3K9me3 peaks and ATAC peaks, and characterized the co-accessible chromatin marks at the opposite ends of H3K9me3-tagged reads, which served as indicators of roles in active transcription. For cross-validation, we intersected the scNanoCT\u0026amp;ATAC-seq H3K9me3 peaks separately with NGS-based ATAC-seq and H3K9me3 ChIP-seq peaks from ENCODE\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cdiv id=\"Sec37\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of active promoters and enhancers\u003c/h2\u003e \u003cp\u003eWe identified active promoters and enhancers across cell types by integrating chromatin states and genomic positions. Specifically, we extracted all ATAC/accessible peaks classified as E1 or E2 states as candidate promoters, and those in the E4 state as candidate enhancers. We applied more stringent criteria for promoter definition: 1) location within \u0026plusmn;\u0026thinsp;1 kb of a TSS of transcripts; 2) when multiple peaks were present near the same TSS, only the peak closest to the TSS were selected; 3) any peak identified as an enhancer in at least one cell type retained the designation to ensure consistent promoter identity across cell types. After promoter annotation, the remaining peaks were classified as enhancers. To validate the reliability of definition, we found the overlap of our promoter/enhancer annotations with the cCREs from ENCODE using BEDtools. We further categorized the promoters/enhancers only detected in GM12878 or K562 as cell-specific and those with genomic coordinate overlaps as common. The statistics and pathway enrichment of the three peak categories were performed as aforementioned.\u003c/p\u003e \u003cdiv id=\"Sec38\" class=\"Section3\"\u003e \u003ch2\u003eIdentification and characterization of E-P pairs\u003c/h2\u003e \u003cp\u003eLeveraging the unique characteristics of scNanoCT\u0026amp;ATAC-seq data: two ends of each read represent co-accessible loci and carry multiple chromatin marks. The number of read pairs spanning two peaks can serve as an indicator of peak co-accessibility. To ensure robustness, only E-P peak pairs supported by at least 10 ATAC/H3K4me3/H3K27ac read pairs were retained for subsequent analysis. Considering the library size selection and sequencing length limit of platform, we calculated the co-accessibility strength of E-P pairs across varied genomic distances, confirming its ability in predicting neighboring co-accessible peaks within the 1\u0026ndash;10 kb range.\u003c/p\u003e \u003cp\u003eFor each cell line, we quantified and normalized the counts of read pairs in E-P pairs, which were sub-classified by cell line specificity, and visualized the clustered co-accessibility strength using the R package ComplexHeatmap\u003csup\u003e\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e v2.14.0. To assess the potential of the E-P pairs as predictor for gene activity, we firstly annotated the heatmap with the difference in co-accessibility strength and the difference in RNA expression for each E-P pair of specific gene. Additionally, we used scatter plots to display their correlation by calculating the Pearson's correlation coefficient (R), associated p-value for significance, as well as the slope and intercept from the linear regression model. For the promoters associated with multiple regulatory enhancers, we grouped them by enhancer count and visualized the corresponding gene expression levels using boxplots. ANOVA test for assessing significant difference and a linear regression fitted to the mean TPM value of groups were performed to illustrate their correlation.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec39\" class=\"Section2\"\u003e \u003ch2\u003eSingle-cell chromatin states analysis on LUAD cells\u003c/h2\u003e \u003cp\u003eThe fundamental single-cell analysis followed the aforementioned procedures, with the following custom analyses. Cells were firstly clustered and annotated based on the ATAC profiles, identifying one cluster of normal epithelial cells (N) and two clusters of tumor cells (T1, T2). Bam files of each cell were used for inferring CNV by Control-FREEC, and fragment files from 3 cell clusters were used for peak calling.\u003c/p\u003e \u003cp\u003eFor each cell type, we broadly categorized H3K9me3 peaks overlapping with ATAC peaks into E2 states and other peaks into E10 states. We identified peaks gained in tumor using thresholds of log2FC\u0026thinsp;\u0026ge;\u0026thinsp;1 and p-value\u0026thinsp;\u0026le;\u0026thinsp;0.05, and peaks loss in tumor with log2FC \u0026le; -1 and p-value\u0026thinsp;\u0026le;\u0026thinsp;0.05. We assigned CNV ratios to peaks based on their location within 1 Mb genomic windows. By comparing the peak fold change (FC\u003csub\u003epeak\u003c/sub\u003e) with the corresponding CNV ratio/fold change (CNV\u003csub\u003epeak\u003c/sub\u003e), between tumor and normal cells, the crosstalk can be summarized into three classes: 1) \u003cem\u003ede novo\u003c/em\u003e increases (FC\u003csub\u003epeak\u003c/sub\u003e - CNV\u003csub\u003epeak\u003c/sub\u003e \u0026ge; 0.5); 2) \u003cem\u003ede novo\u003c/em\u003e decreases (FC\u003csub\u003epeak\u003c/sub\u003e - CNV\u003csub\u003epeak\u003c/sub\u003e \u0026le; -0.5); and 3) passive increases (FC\u003csub\u003epeak\u003c/sub\u003e - CNV\u003csub\u003epeak\u003c/sub\u003e \u0026lt; 0.5 and FC\u003csub\u003epeak\u003c/sub\u003e - CNV\u003csub\u003epeak\u003c/sub\u003e \u0026gt; -0.5) on chromatin marks.\u003c/p\u003e \u003cp\u003eFor survival analysis, we collected the lung adenocarcinoma (LUAD) TCGA datasets from the UCSC Toil RNAseq Recompute Compendium datasets. We utilized the \u0026lsquo;survival\u0026rsquo;\u003csup\u003e61\u003c/sup\u003e package v3.5.5 in R to classify patients into high-expression group (High) and low-expression group (Low) using the best expression cutoff. Then, the survival duration and status of patients in the two groups, Kaplan-Meier curves, log-rank test, and Cox proportional hazards regression on genes were performed to display clinical-related genes.\u003c/p\u003e "},{"header":"Declarations","content":"\u003ch2\u003eAcknowledgments\u003c/h2\u003e \u003cp\u003eOur Flow Cytometry and Sequencing work was performed at the Advanced Cell Technology Core Facility, Guangzhou National Laboratory. This work was supported by grants from the Major Project of Guangzhou National Laboratory (GZNL2024A03001); Shenzhen Medical Research Fund (B2402019); Young Talents Program of Sun Yat-sen University Cancer Center (YTP-SYSUCC-0013); The National Key Research and Development Program Project (2024YFC3406203).\u003c/p\u003e \u003cp\u003e \u003cb\u003eAuthor contributions\u003c/b\u003e \u003c/p\u003e \u003cp\u003eX.F., Z.Z. conceived the project and designed the experiments. D.S. collected the samples and performed library preparation experiments. Z.L. and E.D. did the bioinformatics work. L.L. and Q.H. did the FAM78A validation experiment. J.Z. did the cell culture work. W.X and Z.G. collected the LUAD sample. X. F., Z.Z., D.S. and Z.L. wrote the manuscript. All authors contributed to the discussion and interpretation of the results.\u003c/p\u003e \u003cp\u003e \u003cb\u003eDeclaration of interests\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003ch3\u003eCode availability\u003c/h3\u003e\n\u003cp\u003eCode for processing scNanoCT\u0026amp;ATAC-seq data and Illustrative code snippets for postprocessing are available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/canceromics/scNanoCT-ATAC-seq\u003c/span\u003e\u003cspan address=\"https://github.com/canceromics/scNanoCT-ATAC-seq\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e\u003cdiv id=\"Sec40\" class=\"Section3\"\u003e \u003ch2\u003eData availability\u003c/h2\u003e \u003cp\u003eThe raw sequencing data of the single cells have been deposited to Genome Sequence Archive (GSA) in the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences under accession code PRJCA05098. All other data are available in the article and its Supplementary files .\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSoufi, A. \u003cem\u003eet al.\u003c/em\u003e Pioneer Transcription Factors Target Partial DNA Motifs on Nucleosomes to Initiate Reprogramming. \u003cem\u003eCell\u003c/em\u003e 161, 555\u0026ndash;568 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNicetto, D. \u003cem\u003eet al.\u003c/em\u003e H3K9me3-heterochromatin loss at protein-coding genes enables developmental lineage specification. \u003cem\u003eScience\u003c/em\u003e 363, 294\u0026ndash;297 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, Z. \u003cem\u003eet al.\u003c/em\u003e Structural variants drive context-dependent oncogene activation in cancer. \u003cem\u003eNature\u003c/em\u003e 612, 564\u0026ndash;572 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGeorge, J. \u003cem\u003eet al.\u003c/em\u003e Comprehensive genomic profiles of small cell lung cancer. \u003cem\u003eNature\u003c/em\u003e 524, 47\u0026ndash;53 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarlić, R., Chung, H.-R., Lasserre, J., Vlahoviček, K. \u0026amp; Vingron, M. Histone modification levels are predictive for gene expression. \u003cem\u003eProc. Natl. Acad. Sci. U.S.A.\u003c/em\u003e 107, 2926\u0026ndash;2931 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRao, S., Ahmad, K. \u0026amp; Ramachandran, S. Cooperative binding between distant transcription factors is a hallmark of active enhancers. \u003cem\u003eMolecular Cell\u003c/em\u003e 81, 1651\u0026ndash;1665.e4 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, Z. \u003cem\u003eet al.\u003c/em\u003e Composite transposons with bivalent histone marks function as RNA-dependent enhancers in cell fate regulation. \u003cem\u003eCell\u003c/em\u003e 188, 5878\u0026ndash;5894.e18 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMas, G. \u003cem\u003eet al.\u003c/em\u003e Promoter bivalency favors an open chromatin architecture in embryonic stem cells. \u003cem\u003eNat Genet\u003c/em\u003e 50, 1452\u0026ndash;1462 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBuenrostro, J. D. \u003cem\u003eet al.\u003c/em\u003e Single-cell chromatin accessibility reveals principles of regulatory variation. \u003cem\u003eNature\u003c/em\u003e 523, 486\u0026ndash;490 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCusanovich, D. A. \u003cem\u003eet al.\u003c/em\u003e Multiplex single-cell profiling of chromatin accessibility by combinatorial cellular indexing. \u003cem\u003eScience\u003c/em\u003e 348, 910\u0026ndash;914 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaya-Okur, H. S. \u003cem\u003eet al.\u003c/em\u003e CUT\u0026amp;Tag for efficient epigenomic profiling of small samples and single cells. \u003cem\u003eNat Commun\u003c/em\u003e 10, 1930 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarter, B. \u003cem\u003eet al.\u003c/em\u003e Mapping histone modifications in low cell number and single cells using antibody-guided chromatin tagmentation (ACT-seq). \u003cem\u003eNat Commun\u003c/em\u003e 10, 3747 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBartosovic, M., Kabbe, M. \u0026amp; Castelo-Branco, G. Single-cell CUT\u0026amp;Tag profiles histone modifications and transcription factors in complex tissues. \u003cem\u003eNat Biotechnol\u003c/em\u003e 39, 825\u0026ndash;835 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu, S. J. \u003cem\u003eet al.\u003c/em\u003e Single-cell CUT\u0026amp;Tag analysis of chromatin modifications in differentiation and tumor progression. \u003cem\u003eNat Biotechnol\u003c/em\u003e 39, 819\u0026ndash;824 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, B. \u003cem\u003eet al.\u003c/em\u003e Characterizing cellular heterogeneity in chromatin state with scCUT\u0026amp;Tag-pro. \u003cem\u003eNat Biotechnol\u003c/em\u003e 40, 1220\u0026ndash;1230 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGopalan, S., Wang, Y., Harper, N. W., Garber, M. \u0026amp; Fazzio, T. G. Simultaneous profiling of multiple chromatin proteins in the same cells. \u003cem\u003eMolecular Cell\u003c/em\u003e 81, 4736\u0026ndash;4746.e5 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeers, M. P., Llagas, G., Janssens, D. H., Codomo, C. A. \u0026amp; Henikoff, S. Multifactorial profiling of epigenetic landscapes at single-cell resolution using MulTI-Tag. \u003cem\u003eNat Biotechnol\u003c/em\u003e 41, 708\u0026ndash;716 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBartosovic, M. \u0026amp; Castelo-Branco, G. Multimodal chromatin profiling using nanobody-based single-cell CUT\u0026amp;Tag. \u003cem\u003eNat Biotechnol\u003c/em\u003e 41, 794\u0026ndash;805 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStuart, T. \u003cem\u003eet al.\u003c/em\u003e Nanobody-tethered transposition enables multifactorial chromatin profiling at single-cell resolution. \u003cem\u003eNat Biotechnol\u003c/em\u003e 41, 806\u0026ndash;812 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTedesco, M. \u003cem\u003eet al.\u003c/em\u003e Chromatin Velocity reveals epigenetic dynamics by single-cell profiling of heterochromatin and euchromatin. \u003cem\u003eNat Biotechnol\u003c/em\u003e 40, 235\u0026ndash;244 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu, Y. \u003cem\u003eet al.\u003c/em\u003e scNanoATAC-seq: a long-read single-cell ATAC sequencing method to detect chromatin accessibility and genetic variants simultaneously within an individual cell. \u003cem\u003eCell Res\u003c/em\u003e 33, 83\u0026ndash;86 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, Q. \u003cem\u003eet al.\u003c/em\u003e scNanoSeq-CUT\u0026amp;Tag: a single-cell long-read CUT\u0026amp;Tag sequencing method for efficient chromatin modification profiling within individual cells. \u003cem\u003eNat Methods\u003c/em\u003e 21, 2044\u0026ndash;2057 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe ENCODE Project Consortium \u003cem\u003eet al.\u003c/em\u003e Perspectives on ENCODE. \u003cem\u003eNature\u003c/em\u003e 583, 693\u0026ndash;698 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraun, T. P., Eide, C. A. \u0026amp; Druker, B. J. Response and Resistance to BCR-ABL1-Targeted Therapies. \u003cem\u003eCancer Cell\u003c/em\u003e 37, 530\u0026ndash;542 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eErnst, J. \u0026amp; Kellis, M. Chromatin-state discovery and genome annotation with ChromHMM. \u003cem\u003eNat Protoc\u003c/em\u003e 12, 2478\u0026ndash;2492 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao, Y. \u003cem\u003eet al.\u003c/em\u003e Integrated analysis of multimodal single-cell data. \u003cem\u003eCell\u003c/em\u003e 184, 3573\u0026ndash;3587.e29 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, F., Kan, H. \u0026amp; Castranova, V. Methylation of Lysine 9 of Histone H3. in \u003cem\u003eHandbook of Epigenetics\u003c/em\u003e 149\u0026ndash;157 (Elsevier, 2011). doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/B978-0-12-375709-8.00010-1\u003c/span\u003e\u003cspan address=\"10.1016/B978-0-12-375709-8.00010-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoyt, S. J. \u003cem\u003eet al.\u003c/em\u003e From telomere to telomere: The transcriptional and epigenetic state of human repeat elements. \u003cem\u003eScience\u003c/em\u003e 376, eabk3112 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePehrsson, E. C., Choudhary, M. N. K., Sundaram, V. \u0026amp; Wang, T. The epigenomic landscape of transposable elements across normal human development and anatomy. \u003cem\u003eNat Commun\u003c/em\u003e 10, 5640 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, E. \u003cem\u003eet al.\u003c/em\u003e Landscape of Somatic Retrotransposition in Human Cancers. \u003cem\u003eScience\u003c/em\u003e 337, 967\u0026ndash;971 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, R. \u003cem\u003eet al.\u003c/em\u003e Stage-specific H3K9me3 occupancy ensures retrotransposon silencing in human pre-implantation embryos. \u003cem\u003eCell Stem Cell\u003c/em\u003e 29, 1051\u0026ndash;1066.e8 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLippman, Z. \u003cem\u003eet al.\u003c/em\u003e Role of transposable elements in heterochromatin and epigenetic control. \u003cem\u003eNature\u003c/em\u003e 430, 471\u0026ndash;476 (2004).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLittle, C. D., Nau, M. M., Carney, D. N., Gazdar, A. F. \u0026amp; Minna, J. D. Amplification and expression of the c-myc oncogene in human lung cancer cell lines. \u003cem\u003eNature\u003c/em\u003e 306, 194\u0026ndash;196 (1983).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, X. \u003cem\u003eet al.\u003c/em\u003e Homozygous Deletions and Chromosome Amplifications in Human Lung Carcinomas Revealed by Single Nucleotide Polymorphism Array Analysis. \u003cem\u003eCancer Research\u003c/em\u003e 65, 5561\u0026ndash;5570 (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaykara, O., Bakir, B., Buyru, N., Kaynak, K. \u0026amp; Dalay, N. Amplification of Chromosome 8 Genes in Lung Cancer. \u003cem\u003eJ. Cancer\u003c/em\u003e 6, 270\u0026ndash;275 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu, D., Peters, M., Soltys, V. \u0026amp; Chan, Y. F. Copy number normalization distinguishes differential signals driven by copy number differences in ATAC-seq and ChIP-seq. \u003cem\u003eBMC Genomics\u003c/em\u003e 26, 306 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, L. \u003cem\u003eet al.\u003c/em\u003e ASB7 is a negative regulator of H3K9me3 homeostasis. \u003cem\u003eScience\u003c/em\u003e 389, 309\u0026ndash;316 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins, R. L. \u0026amp; Talkowski, M. E. Diversity and consequences of structural variation in the human genome. \u003cem\u003eNat Rev Genet\u003c/em\u003e 26, 443\u0026ndash;462 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiong, H., Wang, Q., Li, C. C. \u0026amp; He, A. Single-cell joint profiling of multiple epigenetic proteins and gene transcription. \u003cem\u003eSci. Adv.\u003c/em\u003e 10, eadi3664 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCorces, M. R. \u003cem\u003eet al.\u003c/em\u003e An improved ATAC-seq protocol reduces background and enables interrogation of frozen tissues. \u003cem\u003eNat Methods\u003c/em\u003e 14, 959\u0026ndash;962 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. \u003cem\u003eEMBnet j.\u003c/em\u003e 17, 10 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, H. Minimap2: pairwise alignment for nucleotide sequences. \u003cem\u003eBioinformatics\u003c/em\u003e 34, 3094\u0026ndash;3100 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, H. \u003cem\u003eet al.\u003c/em\u003e The Sequence Alignment/Map format and SAMtools. \u003cem\u003eBioinformatics\u003c/em\u003e 25, 2078\u0026ndash;2079 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBolger, A. M., Lohse, M. \u0026amp; Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. \u003cem\u003eBioinformatics\u003c/em\u003e 30, 2114\u0026ndash;2120 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangmead, B. \u0026amp; Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. \u003cem\u003eNat Methods\u003c/em\u003e 9, 357\u0026ndash;359 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQuinlan, A. R. \u0026amp; Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. \u003cem\u003eBioinformatics\u003c/em\u003e 26, 841\u0026ndash;842 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGranja, J. M. \u003cem\u003eet al.\u003c/em\u003e ArchR is a scalable software package for integrative single-cell chromatin accessibility analysis. \u003cem\u003eNat Genet\u003c/em\u003e 53, 403\u0026ndash;411 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, Y. \u003cem\u003eet al.\u003c/em\u003e Model-based Analysis of ChIP-Seq (MACS). \u003cem\u003eGenome Biol\u003c/em\u003e 9, R137 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeers, M. P., Tenenbaum, D. \u0026amp; Henikoff, S. Peak calling by Sparse Enrichment Analysis for CUT\u0026amp;RUN chromatin profiling. \u003cem\u003eEpigenetics \u0026amp; Chromatin\u003c/em\u003e 12, 42 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, S. \u003cem\u003eet al.\u003c/em\u003e Using clusterProfiler to characterize multiomics data. \u003cem\u003eNat Protoc\u003c/em\u003e 19, 3292\u0026ndash;3320 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWagih, O. ggseqlogo: a versatile R package for drawing sequence logos. \u003cem\u003eBioinformatics\u003c/em\u003e 33, 3645\u0026ndash;3647 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRam\u0026iacute;rez, F., D\u0026uuml;ndar, F., Diehl, S., Gr\u0026uuml;ning, B. A. \u0026amp; Manke, T. deepTools: a flexible platform for exploring deep-sequencing data. \u003cem\u003eNucleic Acids Research\u003c/em\u003e 42, W187\u0026ndash;W191 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKent, W. J., Zweig, A. S., Barber, G., Hinrichs, A. S. \u0026amp; Karolchik, D. BigWig and BigBed: enabling browsing of large distributed datasets. \u003cem\u003eBioinformatics\u003c/em\u003e 26, 2204\u0026ndash;2207 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu, G., Wang, L.-G. \u0026amp; He, Q.-Y. ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. \u003cem\u003eBioinformatics\u003c/em\u003e 31, 2382\u0026ndash;2383 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang, T. \u003cem\u003eet al.\u003c/em\u003e Long-read-based human genomic structural variation detection with cuteSV. \u003cem\u003eGenome Biol\u003c/em\u003e 21, 189 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoeva, V. \u003cem\u003eet al.\u003c/em\u003e Control-FREEC: a tool for assessing copy number and allelic content using next-generation sequencing data. \u003cem\u003eBioinformatics\u003c/em\u003e 28, 423\u0026ndash;425 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHaeussler, M. \u003cem\u003eet al.\u003c/em\u003e The UCSC Genome Browser database: 2019 update. \u003cem\u003eNucleic Acids Research\u003c/em\u003e 47, D853\u0026ndash;D858 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePohl, A. \u0026amp; Beato, M. bwtool: a tool for bigWig files. \u003cem\u003eBioinformatics\u003c/em\u003e 30, 1618\u0026ndash;1619 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRao, S. S. P. \u003cem\u003eet al.\u003c/em\u003e A 3D Map of the Human Genome at Kilobase Resolution Reveals Principles of Chromatin Looping. \u003cem\u003eCell\u003c/em\u003e 159, 1665\u0026ndash;1680 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGu, Z. Complex heatmap visualization. \u003cem\u003eiMeta\u003c/em\u003e 1, e43 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTherneau, T. M. \u0026amp; Grambsch, P. M. The Cox Model. in \u003cem\u003eModeling Survival Data: Extending the Cox Model\u003c/em\u003e 39\u0026ndash;77 (Springer New York, New York, NY, 2000). doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/978-1-4757-3294-8_3\u003c/span\u003e\u003cspan address=\"10.1007/978-1-4757-3294-8_3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8125184/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8125184/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eProfiling multiple histone modifications alongside chromatin accessibility at single-cell resolution has been enabled by technologies such as multi-CUT\u0026amp;Tag, scGET-seq, and nano-CUT\u0026amp;Tag. Here, we developed Nanopore sequencing based single-cell CUT\u0026amp;Tag and ATAC co-profiling (scNanoCT\u0026amp;ATAC-seq), which allows for the simultaneous mapping of genome mutations, histone modifications and chromatin accessibility within the same individual cells. scNanoCT\u0026amp;ATAC-seq achieves comparable sensitivity and fragment yield per cell as previous single-modal methods. Anchoring by the ATAC module, a comprehensive epimap that includes multiple histone marks is characterized, generating a cell type specific annotation of chromatin status for genomic regulatory elements, which specified different H3K9me3 modes and identified enhancer-promoter interactions at single-molecular level, specifying genomic structure variations lead to novel regulatory elements in cancer cells. Applying to lung adenocarcinoma, scNanoCT\u0026amp;ATAC-seq clearly captured the tumor subclones and revealed the coordination of genomic and epigenetic regulations. Together, scNanoCT\u0026amp;ATAC-seq reveals the interplay between genomic and multiple epigenetic modalities underlying any cellular processes.\u003c/p\u003e","manuscriptTitle":"Parallel profiling of genome mutations and multiple chromatin modalities in single cells by scNanoCT\u0026amp;ATAC-seq","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-13 15:10:04","doi":"10.21203/rs.3.rs-8125184/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"498db91f-1602-4051-aaaf-02825bb1c7b4","owner":[],"postedDate":"January 13th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":60782822,"name":"Biological sciences/Biotechnology/Sequencing"},{"id":60782823,"name":"Biological sciences/Genetics/Epigenomics"}],"tags":[],"updatedAt":"2026-04-03T02:15:52+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-13 15:10:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8125184","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8125184","identity":"rs-8125184","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}