Integrated transcriptomic and machine learning-driven analysis reveals high-confidence circular RNA biomarkers in Lung Adenocarcinoma

Integrated transcriptomic and machine learning-driven analysis reveals high-confidence circular RNA biomarkers in Lung Adenocarcinoma | 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 Research Article Integrated transcriptomic and machine learning-driven analysis reveals high-confidence circular RNA biomarkers in Lung Adenocarcinoma Ayushi Malviya, Rajabrata Bhuyan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8684357/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 Lung cancer remains the leading cause of cancer-related deaths worldwide, with lung adenocarcinoma (LUAD) as its most prevalent subtype. Circular RNAs (circRNAs), known for acting as microRNA sponges and interacting with RNA-binding proteins, have emerged as key regulators in cancer biology. In this study, we introduce an integrated framework combining transcriptome profiling, network analysis, and machine learning to systematically identify and prioritise potential circRNA biomarkers in LUAD. We analysed RNA-seq datasets from LUAD samples and identified 52,744 circRNAs (18,922 novel and 33,822 previously known). Differential expression analysis revealed 1,480 significantly dysregulated circRNAs (855 downregulated and 625 upregulated) between tumour and normal tissues. To overcome limitations of traditional single-parameter screening, we implemented a three-pronged machine learning strategy integrating feature-weighted statistical scoring, unsupervised clustering with outlier detection, and deep neural networks. This multi-algorithm consensus approach identified 34 high-confidence circRNAs (17 upregulated and 17 downregulated) consistently prioritised across all methods, greatly reducing false positives. Functional enrichment analyses revealed distinct roles: upregulated circRNAs predominantly orchestrate metabolic reprogramming, epithelial–mesenchymal transition, and cytoskeletal remodelling, whereas downregulated circRNAs govern transcriptional control, apoptosis regulation, and tumour-suppressor pathways. Key biomarker candidates include has_circ_0024109, hsa_circ_0058736, hsa_circ_0052195, hsa_circ_0085740, and hsa_circ_0060931 (upregulated), and hsa_circ_0016392, hsa_circ_0058622, hsa_circ_0012248, hsa_circ_0037308, and hsa_circ_0048053 (downregulated). This study provides a comprehensive, machine-learning-validated circRNA biomarker panel for LUAD, offering mechanistic insights into circRNA-driven oncogenesis and laying a foundation for next-generation circRNA-based diagnostics and therapeutics in precision oncology. circRNAs LUAD Machine Learning Biomarkers Precision Oncology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Lung cancer is responsible for over 1.8 million deaths each year, continues to pose a significant global health risk, and is recognised as the primary cause of cancer-related fatalities worldwide [ 1 ]. Its five-year survival rate remains extremely low despite developments in diagnosis and treatment, especially for non-small cell lung cancer (NSCLC), which accounts for approximately 85% of cases [ 2 ]. Lung adenocarcinoma, the most common subtype of NSCLC, has a consistently poor prognosis, and recurrent resistance to targeted therapy [ 3 ]. The molecular mechanisms underlying lung cancer involve complex regulatory networks including genetic mutations, epigenetic alterations, and dysregulated non-coding RNAs, among which circRNAs have emerged as critical modulators of tumour initiation, progression, and drug resistance [ 4 ]. Circular RNAs (circRNAs) are generated through alternative splicing events that produce covalently closed loop structures, conferring remarkable resistance to exonuclease-mediated degradation, and thereby enhancing their stability [ 5 ], [ 6 ], [ 7 ]. These molecules frequently display tissue-specific and developmentally regulated expression patterns [ 8 ]. Initially viewed as transcriptional artifacts, they are now recognized as multifunctional regulators that modulate gene expression by sponging microRNAs, binding RNA-binding proteins (RBPs), and in some cases, facilitating cap-independent translation of bioactive peptides. This evolving understanding has positioned circRNAs as promising biomarkers and potential therapeutic targets in various malignancies [ 9 ]. While several circRNAs have been implicated in lung cancer progression through modulation of proliferation, invasion, metastasis, and drug resistance, our understanding of their full functional spectrum remains incomplete. This study presents a comprehensive exploration of the functional landscape of circRNAs in lung cancer, integrating multi-level analyses of circRNA-miRNA-mRNA regulatory networks, and RBP interactions. By leveraging state-of-the-art AI/ML approaches, we prioritized statistically significant and biologically functional circRNA candidates and predicted their biomarker potential. The aim of the study is to elucidate the new regulatory roles of circRNAs establishing a basis for the next generation of circRNA-based diagnostics and therapies, offering insights poised to transform the lung cancer research. Materials and Methods Dataset collection The NCBI Sequence Read Archive (SRA) was systematically explored to retrieve publicly available RNA-Seq datasets optimized for circRNA detection [ 10 ]. Only transcriptomic paired end reads from LUAD samples enriched for non-coding RNAs were considered. The selected datasets were categorized based on their intended application: ncRNA-enriched libraries were retained for circRNA identification (Accession IDs: SRP335547; SRP089923; SRP048484; SRP047399), while libraries containing both cancerous and non-cancerous samples were utilized for differential expression analyses. All RNA-Seq libraries were obtained from the European Nucleotide Archive [ 11 ]. The human reference genome (GRCh37.fa) along with its corresponding annotation file (ucsc_GRCh37.gtf) was downloaded from the UCSC Genome Browser [ 12 ]. For annotation of identified circRNAs, reference files from the CircBase database (hsa_hg19_circRNA.bed; hsa_hg19_circRNA.gtf) were employed [ 13 ]. In addition, mature and annotated miRNA sequences were retrieved from the miRBase database ( homo_sapiens.fa) [ 14 ]. CircRNA detection Pre-processing and quality control of each dataset were conducted using FastQC { http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ } to assess read quality. Reads with a base quality score below Q20 were dropped and the sequences were trimmed using Trimmomatic [ 15 ]. The quality-filtered reads were then subjected to circRNA detection using CIRI2, the updated version of the CIRI (circRNA identifier) algorithm, which relies on chiastic clipping signals and multiple filtration steps to reliably and unbiasedly detect circRNAs [ 16 ]. It utilises BWA for spliced alignment against the human reference genome [ 17 ]. CIRI2 was chosen for this study because it offers a strong balance of accuracy, computational efficiency, and low resource demand. CircRNA detection was performed individually for each sample, and the resulting output files were subsequently merged into a consolidated BED file for downstream analyses. Differential gene expression studies StringTie [ 18 ] was used for transcript normalization, abundance estimation, and generation of count data for downstream analysis using the Bioconductor package DESeq2 [ 19 ]. Differential expression analysis was performed using TPM-normalized count values. CircRNA transcripts from cancer and normal samples were further filtered based on an adjusted p-value threshold of 0.05, followed by refinement using a log 2 fold change cut-off of ± 2. Prediction of target miRNAs and RBPs The interactions between significantly differentially expressed circRNAs and their target miRNAs were predicted using the miRanda tool [ 20 ], applying a threshold score of 150 and a minimum free energy cutoff of − 25 to ensure the inclusion of a sufficiently comprehensive pool for feature extraction required in the next steps. An interaction network comprising the top-ranking circRNAs and their high-confidence miRNA targets was subsequently constructed and visualized using Cytoscape [ 21 ]. Simultaneously, target RNA-binding proteins (RBPs) of the significantly dysregulated circRNAs were retrieved from the circInteractome database [ 22 ], and circRNA-RBP interaction networks were generated in Cytoscape for both upregulated and downregulated circRNAs. Machine Learning-mediated biomarker potential prediction A circRNA feature table was constructed containing six biologically relevant attributes: Exclusivity Score, Consistency Score, Differential Expression Score (DE_Score), Binding Affinity, Regulatory Potential, and Sponge Capacity Score ( Supplementary Table 1) . This multi-dimensional dataset provides a comprehensive representation of circRNA functionality and biomarker potential. Three complementary computational frameworks were applied on this feature table. First, a feature-weighted statistical scoring approach generated a Biomarker Potential Score through the integration of weighted features, yielding an interpretable baseline ranking. Second, unsupervised machine learning (PCA, K-means clustering, and Isolation Forest) was used to identify intrinsic patterns, stratify circRNAs into high/medium/low groups, and highlight potential outliers. A normalized, weighted biomarker score was calculated within these clusters. Third, a deep learning model was implemented using a multilayer neural network, operating in supervised or semi-supervised mode, to capture nonlinear feature interactions and assign probabilistic scores of biomarker relevance. Consensus circRNAs were defined as those consistently prioritized by all three methods. This integrative, multi-tiered strategy reduced false positives, enhanced reproducibility, and outperformed traditional single-parameter screening by ensuring a more robust and biologically meaningful selection of circRNA biomarker candidates. Gene Ontology and pathway enrichment studies To identify the miRNA-targeted genes for subsequent GO analysis, the multimiR [ 23 ] package in R was employed. Three target prediction databases were queried (miRAnda, mirbase and TargetScan) [ 14 ], [ 20 ], [ 24 ], and only those genes that were predicted by at least two of the three databases and/or were experimentally validated were included in the final list. Functional annotation of significantly dysregulated circRNA target genes, identified through their interacting miRNAs, was performed using Gene Ontology (GO) analysis via the Enrichr server [ 25 ]. Gene functions were categorized into three domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). These categories were separately analyzed for the upregulated and downregulated gene sets. Additionally, pathway enrichment analysis was conducted using the KEGG database [ 26 ] . Results Characterization of circRNA expression and host gene associations . The circRNA profile in LUAD was detected using the CIRIv2 tool and subsequently annotated with circBase database. CIRI2 employs maximum likelihood-based multiple seed matching to effectively distinguish true back-splice junction (BSJ) reads from incorrectly mapped non-BSJ reads, enabling both low false-discovery rates and high sensitivity. Its enhanced improved read-classification accuracy, combined with multithreading support, allows for faster execution and more efficient RAM usage. This systematic and rigorous workflow facilitated the accurate identification, distinction, and classification of circRNAs using the circbase database, allowing them to be organized into two primary categories: previously unreported novel circRNAs and those already catalogued as known circRNAs ( Fig. 1 a ) . “Novel circRNAs” are defined as circular RNAs identified in LUAD samples that do not overlap with, or correspond to, any circRNAs previously annotated in the circBase database, and thus represent newly identified and uncharacterized circRNA transcripts. Conversely, “known circRNAs” are those whose genomic coordinates and features match entries previously documented in circBase, representing validated or annotated circular RNA species. In the LUAD dataset, 18,922 circRNAs were classified as Novel, while a larger number, 33,822 circRNAs, were identified as Known. The majority of circRNAs, both novel and known, mapped predominantly to exonic regions (32,452), consistent with current understanding that exonic circRNAs constitute the bulk of circRNA populations in cancer ( Fig. 1 b ) . A smaller number of circRNAs derived from intronic regions (1,194), reflecting the diversity of circRNA biogenesis beyond exons. Additionally, 2,656 circRNAs were categorized as overlapping, spanning multiple genomic features and indicating complex regulatory loci that may be hotbeds for alternative circularization or transcript overlap. Non-coding circRNAs totalled 1,199, highlighting the importance of circRNAs beyond protein-coding potential in regulatory networks. A minor subset of 86 antisense circRNAs was identified, indicating specific regulatory roles potentially linked to gene expression interference or silencing. Differential expression of circRNAs in LUAD vs normal lung cells Differential gene expression analysis revealed substantial transcriptomic alterations in lung adenocarcinoma compared to normal lung tissue. A total of 1,480 circRNAs exhibited significant differential expression (adjusted p-value < 0.05), with 855 circRNAs demonstrating downregulation and 625 circRNAs showing upregulation in tumour samples ( Fig. 2 b ) . This pattern, characterized by a greater proportion of downregulation (57.8%) relative to their higher expression (42.2%), suggests that lung adenocarcinoma pathogenesis involves extensive transcriptional repression alongside oncogenic activation. Complementary insights into the expression patterns and sample clustering for circRNAs in LUAD study are offered by the accompanying heatmap and PCA plots ( Fig. 2 a, Fig. 2 c ) . While the PCA plot, which is based on principal component analysis of circRNA expression profiles, visually distinguishes between cancer and normal samples, the heatmap shows the relative expression levels of certain circRNAs across different samples. The predominance of downregulated circRNAs may reflect the loss of normal lung epithelial cell identity and differentiation programs during malignant transformation. Conversely, the upregulated circRNAs likely encompass oncogenic drivers, genes involved in proliferation, metabolic reprogramming, and immune evasion mechanisms characteristic of lung cancer. The bilateral nature of these expression changes, with substantial alterations in both directions, highlights the complex rewiring of cellular transcriptional networks that accompanies tumorigenesis and supports the multifactorial nature of lung adenocarcinoma development. CircRNA-miRNA-mRNA regulatory network and RBP interactions Analysis of the upregulated circRNAs revealed a distinct interaction network ( Fig. 3 a ) characterized by a smaller but relatively focused set of circRNAs that engage with specific miRNAs. Prominent circRNAs included hsa_circ_0012003, hsa_circ_0026425, hsa_circ_0027924, hsa_circ_0035015, hsa_circ_0043614, hsa_circ_0043632, hsa_circ_0044130, hsa_circ_0052195, hsa_circ_0085288, and hsa_circ_0085740. Several of these circRNAs acted as hub molecules by interacting with multiple miRNAs; for example, hsa_circ_0035015 (targeting hsa-miR-665, hsa-miR-770-5p, hsa-miR-4739), hsa_circ_0043614 (targeting hsa-miR-7158-5p, hsa-miR-6894-5p), and hsa_circ_0043632 (targeting hsa-miR-4776-3p, hsa-miR-7106-5p, hsa-miR-1293, hsa-miR-6791-3p). This pattern suggests that a few highly expressed circRNAs may exert substantial influence over a diverse repertoire of miRNAs. Network analysis of the downregulated circRNAs ( Fig. 3 B ) revealed a complex regulatory catalogue comprising several circRNAs, notably hsa_circ_0049036, hsa_circ_0046727, hsa_circ_0024652, hsa_circ_0092361, hsa_circ_0002734, hsa_circ_0012248, hsa_circ_0084884, hsa_circ_0000245, hsa_circ_0011819, hsa_circ_0082722, and hsa_circ_0079823, that interact with a wide spectrum of miRNAs, including hsa-miR-4632-5p, hsa-miR-330-5p, hsa-miR-3714, hsa-miR-519a-2-5p, hsa-miR-6088, hsa-miR-615-5p,hsa-miR-615-5p, hsa-miR-197-5p, hsa-miR-612, hsa-miR-15a-3p, and hsa-miR-324-3p, among others. Several circRNAs exhibited hub-like properties by targeting multiple miRNAs, suggesting that they may exert broad regulatory influence. Conversely, certain miRNAs such as hsa-miR-197-5p, hsa-miR-1285-3p, hsa-miR-619-5p, hsa-miR-363-5p, hsa-miR-7110-5p, and hsa-miR-328-5p were recurrently targeted by multiple circRNAs, indicating their central role in the post-transcriptional network. The circRNA-RBP interaction network ( Fig. 4 ) in lung cancer reveals a complex regulatory architecture involving 27 differentially expressed circRNAs (13 upregulated, 14 downregulated) and 24 RNA-binding proteins exhibiting distinct interaction patterns. Network topology analysis demonstrates that upregulated circRNAs display significantly higher RBP connectivity compared to their downregulated counterparts, with hsa_circ_0005352 emerging as a central hub interacting with 17 RBPs, followed by hsa_circ_0085740 (10 RBPs) and hsa_circ_0044130 (6 RBPs). Among the RBPs identified, EIF4A3 functions as the most promiscuous binding partner, interacting with 22 circRNAs from both expression cohorts, while AGO2 targets 15 circRNAs and HuR binds 11 circRNAs, establishing these proteins as convergent regulatory hubs that bridge upregulated and downregulated circRNA populations. The IGF2BP protein family (IGF2BP1/2/3) demonstrates preferential binding to upregulated circRNAs, particularly hsa_circ_0005352, hsa_circ_0044130, and hsa_circ_0085740, suggesting their role in stabilizing oncogenic circRNA transcripts. Notably, METTL3 and DGCR8 exclusively interact with downregulated circRNAs, while C22ORF28, EWSR1, TAF15, and specific members of the LIN28 and FXR families show selective binding to upregulated circRNAs. Splicing factors U2AF65 and ZC3H7B uniquely associate with downregulated circRNAs, whereas stress granule components CAPRIN1 and FXR1/2 exhibit differential targeting patterns. The extensive interconnectivity of upregulated circRNAs with multiple RBPs suggests enhanced post-transcriptional stability mechanisms driving their accumulation in lung cancer cells Machine Learning driven biomarker prediction We identified 34 circular RNA (circRNA) candidates as high-potential biomarkers through a comprehensive multi-method consensus ranking approach that integrated three independent analytical frameworks: deep learning-based analysis (DL score), clustering-based assessment (Cluster score), and statistical calculation (Stats score). All candidates achieved DL scores exceeding 0.987, demonstrating robust discriminatory potential and were derived from genes with diverse biological functions. Each approach highlights different aspects of biological significance: the statistical model emphasizes disease specificity, consistency, and regulatory or sponge capacity; the unsupervised methods reveal circRNAs central to meaningful biological clusters or with high-impact outlier profiles; and the deep learning model captures nonlinear feature interactions predictive of disease relevance (Supplementary Table 1) . While reducing false positives and boosting confidence in their possible therapeutic efficacy, the convergence of these orthogonal approaches on the same set of circRNAs highlights their multifaceted strengths, such as being structurally pivotal in regulatory networks, reproducible across datasets, and biologically distinctive. Among upregulated candidates, hsa_circ_0085740 from PTK2 (focal adhesion kinase, LFC: 10.6) demonstrated exceptional performance with top 4 rankings in clustering and statistical analyses. The hsa_circ_0060931 from CYP24A1, a key enzyme in vitamin D metabolism, exhibited the highest fold-change, while hsa_circ_0052195 from PTPRH (protein tyrosine phosphatase receptor, LFC: 8.97) maintained consistent top 10 rankings across all methods. Other key upregulated candidates included hsa_circ_0058736 from C2orf82, hsa_circ_0024109 from MMP1 (matrix metalloproteinase involved in extracellular matrix remodelling), and hsa_circ_0043632 from KRT17 (keratin 17, a cytoskeletal protein).The top-ranked downregulated circRNAs included hsa_circ_0058622 from the SP100 gene, which exhibited exceptional consensus with rank 1 across all three methods, followed by hsa_circ_0012248 from NASP gene and hsa_circ_0037308 from TELO2 gene, both showing consistent top-5 rankings. Notably, hsa_circ_0016392 derived from PPP2R5A, a regulatory subunit of protein phosphatase 2A, displayed the most extreme downregulation (LFC: -13.21) with strong statistical support. Additional downregulated candidates of interest included hsa_circ_0048053 from SHC2 (adaptor protein in signalling pathways), hsa_circ_0046727 from SMCHD1 (chromatin modifier), and hsa_circ_0003663 from TOP3A (DNA topoisomerase), all demonstrating coordinated top 10 rankings. The top 34 dysregulated circRNAs (17 upregulated and 17 downregulated) ( Table 1 ) were subsequently selected as the final high-confidence biomarker panel for downstream functional validation, including Gene Ontology (GO) term analyses and pathway enrichment studies, to investigate their roles in lung cancer biology. The enrichment results provided independent evidence linking these circRNAs to processes such as transcriptional regulation, EMT, metabolic adaptation, and invasive signalling, further reinforcing their relevance to lung cancer progression. This integrative workflow demonstrates that multi-algorithm consensus combined with biological pathway validation strengthens biomarker discovery. Table 1 List of top-scoring circRNAs identified across three independent ranking algorithms circRNA Gene name Expression LFC value DL score DL rank Cluster score Cluster rank Stats score Stats rank hsa_circ_0058622 SP100 Low -12.17 0.99663204 1 0.757 1 0.740 1 hsa_circ_0012248 NASP Low -11.18 0.99612737 2 0.744 3 0.727 2 hsa_circ_0037308 TELO2 Low -9.94 0.9952877 3 0.728 5 0.71 4 hsa_circ_0000245 None Low -2.56 0.9949515 4 0.608 14 0.650 13 hsa_circ_0085751 PTK2 High 4.94 0.99449205 5 0.558 25 0.643 16 hsa_circ_0048053 SHC2 Low -9.48 0.99448025 6 0.716 6 0.700 5 hsa_circ_0046727 SMCHD1 Low -9.41 0.99440217 7 0.713 7 0.696 6 hsa_circ_0052195 PTPRH High 8.97 0.9940776 8 0.707 8 0.695 7 hsa_circ_0058736 C2orf82 High 8.27 0.99389577 9 0.587 18 0.671 9 hsa_circ_0085740 PTK2 High 10.6 0.99388254 10 0.733 4 0.716 3 hsa_circ_0084884 TMEM67 Low -5.89 0.99361026 11 0.572 22 0.655 12 hsa_circ_0003663 TOP3A Low -8.2 0.9934958 12 0.696 9 0.679 8 hsa_circ_0092334 PRELID2 High 5.03 0.9927882 13 0.537 33 0.630 23 hsa_circ_0005352 CDCA2 High 5.34 0.9924915 14 0.539 30 0.629 24 hsa_circ_0024109 MMP1 High 6.72 0.99242914 15 0.575 21 0.659 11 hsa_circ_0011819 MACF1 Low -3.45 0.9921859 16 0.636 13 0.631 22 hsa_circ_0024652 GRIK4 Low -4 0.9918901 17 0.538 31 0.624 26 hsa_circ_0060931 CYP24A1 High 28.43 0.9917231 18 0.579 20 0.645 15 hsa_circ_0085288 CTHRC1 High 4.46 0.9914436 19 0.538 32 0.636 19 hsa_circ_0043632 KRT17 High 5.21 0.99133587 20 0.553 27 0.643 17 hsa_circ_0092004 FLNA Low -3.98 0.99129635 21 0.517 39 0.612 30 hsa_circ_0016392 PPP2R5A Low -13.21 0.991119 22 0.690 10 0.639 18 hsa_circ_0092361 VIPR1 Low -4.84 0.99085486 24 0.545 29 0.635 20 hsa_circ_0026425 KRT6A High 4.97 0.99053043 25 0.548 27 0.634 21 hsa_circ_0079823 BMPER Low -4.75 0.9904053 27 0.553 26 0.647 14 hsa_circ_0079251 SLC29A4 High 4.12 0.9901022 29 0.503 42 0.602 40 hsa_circ_0044130 KIF18B High 4.07 0.9897237 30 0.508 41 0.606 33 hsa_circ_0002734 SEMA5A Low -2.94 0.98954356 32 0.669 11 0.627 25 hsa_circ_0035015 CKMT1B High 4.33 0.9894872 33 0.525 37 0.618 29 hsa_circ_0027924 C12orf48 High 3.66 0.98910236 36 0.499 49 0.599 49 hsa_circ_0049036 FBN3 Low -3.65 0.98865277 37 0.5117 40 0.605 34 hsa_circ_0012003 SLC2A1 High 3.35 0.9880233 40 0.497 50 0.603 37 hsa_circ_0043614 KRT14 High 4.01 0.98796755 41 0.526 36 0.619 27 hsa_circ_0082722 DENND2A Low -2.81 0.9872055 49 0.501 46 0.598 46 DL score : Deep learning-based score; Cluster score : Clustering based score; Stats scores : Statistically calculated scores; All three represent independent ranking methods; LFC shows expression fold change. Target gene set enrichment analysis The target genes of circRNA-miRNA regulatory axis were subjected to Gene Ontology (GO) studies aimed at characterizing the biological processes underlying mechanism of Lung Cancer progression, to provide valuable insights into their pathogenesis. The Gene functions were systematically categorized into the three major Gene Ontology (GO) domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). To capture directional biological changes, GO enrichment analyses were performed separately for the upregulated and downregulated gene sets, enabling the identification of distinct functional regulation associated with each expression pattern. In parallel, pathway enrichment analysis was conducted using the KEGG database. This integrated GO-KEGG framework enabled a comprehensive investigation of the functional landscape, facilitating the identification of key pathways, cellular modules, and molecular mechanisms potentially driving disease progression . The GO enrichment analysis of genes associated with upregulated circRNAs revealed strong activation of pathways linked to transcriptional activity, epithelial remodelling, and membrane-associated processes ( Supplementary Fig. 1 ). An elevation in overall transcriptional activity was observed through key BP terms enrichment, including DNA-templated transcription, RNA biosynthetic processes, and RNA polymerase II–mediated transcription. Enrichment of epithelium development, positive regulation of epithelial cell proliferation, and cell-cell adhesion, with enhanced proliferative and structural dynamics within epithelial cells were observed. Additionally, prominent terms such as neuron projection development and projection morphogenesis, suggesting cytoskeletal and structural reorganization, and cellular plasticity were inferred. The CC terms associated with upregulated genes mapped strongly to the recycling endosome, exocytic vesicle membrane, synaptic vesicle membrane, and cortical endoplasmic reticulum, that could suggest increased vesicle trafficking and membrane turnover. The presence of motile cilium and peroxisome further highlights involvement in signalling and metabolic microdomains. The MF analysis revealed enrichment in transporter activities (amino acid transporters, metal cation antiporters) and regulatory elements such as kinase activator activity, PDZ domain binding, and actin binding. The simultaneous activation of proliferative and transport-related pathways, alongside cytoskeletal reorganization, could be suggestive of upregulated circRNA networks driving growth-oriented and invasive cellular programs in LUAD. The GO analysis of downregulated circRNA-associated genes showed a contrasting enrichment pattern dominated by processes such as dual regulation of DNA-templated transcription, gene expression regulation, RNA polymerase II regulation, regulation of apoptotic processes, and intracellular signalling moderation, which could be indicative of attenuation of multiple regulatory checkpoints governing transcription and cell death ( Supplementary Fig. 2 ). Suppression of pathways controlling cell population proliferation and intracellular signal transduction suggests reduced regulatory constraint on growth and signalling fidelity. The key enriched CC terms included focal adhesion, cell substrate junction, ER membrane, mitochondrial membrane, and nuclear lumen, highlighting downregulation within adhesion structures, energy-regulating organelles, and nuclear regulatory spaces. In the MF domain, key downregulated terms involved DNA-binding transcription activator activity, cis-regulatory region binding, kinase binding, serine/threonine kinase activity, and GTPase regulator activity. All of which could be suggestive of decreased negative regulation of kinase pathways and transcriptional control mechanisms. Enrichment of significant pathways A comparative KEGG pathway analysis of target genes regulated by dysregulated circRNA-miRNA interaction revealed a distinct division of labour between upregulated and downregulated circRNAs-miRNA networks, which coordinate unique but complementary activities that might contribute to cancer progression. Upregulated circRNAs showed predominant associations with several disease- and metabolism-related pathways. Enriched pathways included Herpes simplex virus 1 infection, pathogenic E. coli infection, and Shigellosis, indicating activation of infection-responsive signalling modules. Additional enrichment was observed in pathways related to fluid shear stress and atherosclerosis, cellular senescence, gastric cancer, breast cancer, and the TGF-β signalling pathway, suggesting involvement of stress responses and cancer-linked signalling cascades. Metabolic pathways such as glycolysis/gluconeogenesis, RNA degradation, and amino-acid metabolism (glycine-serine-threonine metabolism and phenylalanine metabolism) were also significantly overrepresented. Pathways related to cytoskeletal regulation, including regulation of actin cytoskeleton, appeared among the enriched terms as well. Together, these enriched pathways indicate that upregulated circRNA-associated genes may support LUAD progression by enhancing metabolic activity, stress adaptation, and oncogenic signalling programs. In contrast, downregulated circRNAs showed active associations with classical oncogenic signalling cascades representing of multiple signalling, cancer-associated, and hormonal regulation pathways. Enriched pathways included the cAMP signalling pathway, cGMP-PKG signalling pathway, and the MAPK signalling pathway, indicating reduced activity across major intracellular signalling routes. Several cancer-linked pathways were present, including pathways in cancer, proteoglycans in cancer, colorectal cancer, and renal cell carcinoma. Additional downregulated pathways involved ErbB signalling, adherens junction, and axon guidance, reflecting reduced signalling coordination and structural regulation. Hormonal and endocrine-associated pathways such as aldosterone synthesis and secretion, growth hormone synthesis, secretion and action, and the longevity regulating pathway also appeared among the enriched terms. Viral response pathways, including human cytomegalovirus infection, were similarly represented. The collectively downregulated pathways signify diminished signalling fidelity, compromised structural and adhesion mechanisms, and the suppression of regulatory cancer-associated modules, which may jointly facilitate LUAD progression by undermining cellular control systems that typically inhibit malignant behaviour. Discussion Mounting evidence have suggested that circRNAs, with their remarkable stability and regulatory versatility, predictive and diagnostic biomarkers for various forms of cancers [ 27 ]. Yet, the exact intricate interplay between dysregulated circRNAs, their target miRNAs, and downstream gene networks remains largely ambiguous, leaving a critical gap in our comprehension of how these molecules shape tumour behaviour. Addressing this gap requires not only molecular characterization but also rigorous biomarker validation. This study offers an extensive and multidimensional exploration of the circRNA regulatory landscape in lung adenocarcinoma, revealing how deeply these non-coding transcripts are interwoven into the molecular architecture of tumour development and progression. CircRNAs have long remained an underexplored component of cancer biology, yet their exceptional stability, tissue-specific expression, and intricate interactions with miRNAs and RBPs position them as crucial regulators rather than mere transcriptional by-products. Our findings reinforce this emerging paradigm by demonstrating that circRNA dysregulation in LUAD is strongly associated with widespread alterations in transcriptional activity, signalling dynamics, cellular remodelling, and metabolic adaptation. Using a systematic workflow that combined differential expression profiling, circRNA-miRNA-mRNA network reconstruction, RBP interaction prediction, and functional enrichment analysis, we identified a diverse set of circRNAs exhibiting significant dysregulation between cancer vs normal tissues. To strengthen the reliability of these molecular insights, we employed a three-layered biomarker validation strategy that integrated statistical testing, unsupervised clustering, and supervised machine-learning based prediction. This triangulated approach enabled the identification of 34 robust circRNA candidates consistently validated across all methodological frameworks ( Table 1 ) . Unlike single-method analyses that often produce context-dependent or noisy biomarker lists, this consensus strategy substantially enhances diagnostic confidence and translational potential. The global circRNA profiling of LUAD revealed a broad and structurally diverse circular transcriptome, comprising both known and novel circRNAs, with exon-derived species forming the predominant class. This distribution aligns with established circRNA biogenesis patterns in cancer, while the presence of intronic, overlapping, non-coding, and antisense circRNAs highlights additional regulatory layers that may be disrupted during malignant transformation. Differential expression analysis demonstrated substantial dysregulation, marked by a higher proportion of downregulated circRNAs compared to upregulated ones. This trend suggests widespread loss of epithelial identity and transcriptional repression in tumour tissue, whereas the upregulated subset likely represents circRNAs that support proliferative and oncogenic programs. The stark contrast between tumour and normal samples in PCA and heatmap analyses further highlight the distinct circRNA expression landscape characteristic of LUAD. Several of the top dysregulated circRNAs originated from genes involved in focal adhesion dynamics, extracellular matrix remodelling, cytoskeletal regulation, metabolic pathways, and chromatin or phosphatase-mediated control, biological processes well established in cancer progression. Notably, multiple circRNAs reported here, including those derived from PTK2, MMP1, KRT17, and CYP24A1 have been previously implicated in other malignancies, supporting their broader relevance as tumour-associated circular transcripts [ 28 ], [ 29 ], [ 30 ], [ 31 ], [ 32 ], [ 33 ], [ 34 ]. This recurrence across cancer types strengthens the likelihood that these circRNAs serve conserved roles in growth signalling, invasion, and stress adaptation, and other fundamental cancer hallmarks [ 35 ], [ 36 ] . Analysis of the miRNA interaction networks revealed distinct regulatory architectures associated with upregulated and downregulated circRNAs in LUAD. The upregulated circRNAs formed a relatively compact but strategically focused network, with circRNAs such as hsa_circ_0035015, hsa_circ_0043614, and hsa_circ_0043632 functioning as prominent hubs by engaging multiple miRNAs. This concentrated interaction pattern suggests that a limited subset of highly expressed circRNAs may exert disproportionate regulatory influence. Conversely, the downregulated circRNAs constituted a more expansive and intricate network. Key circRNAs, including hsa_circ_0049036, hsa_circ_0046727, hsa_circ_0012248, and hsa_circ_0092361, interacted with a broad repertoire of miRNAs, indicating wider post-transcriptional disruption. While several miRNAs, e.g., hsa-miR-197-5p, hsa-miR-1285-3p, and hsa-miR-619-5p were observed to be recurrently targeted by multiple circRNAs, underscoring their centrality within the regulatory landscape. Collectively, these patterns reflect differential modes of network modulation associated with circRNA dysregulation in LUAD. The circRNA-RBP interaction network in LUAD reveals a highly structured regulatory architecture involving 27 dysregulated circRNAs and 24 RNA-binding proteins with distinct binding patterns. Upregulated circRNAs exhibited substantially higher RBP connectivity, with hsa_circ_0005352 emerging as the principal hub interacting with 17 RBPs, followed by hsa_circ_0085740 and hsa_circ_0044130. Among RBPs, EIF4A3 displayed the most extensive binding spectrum, engaging 22 circRNAs across both expression groups, while AGO2 and HuR also interacted broadly, underscoring their roles as central regulators within the post-transcriptional network. The IGF2BP family demonstrated selective affinity toward upregulated circRNAs, particularly hsa_circ_0005352, hsa_circ_0044130, and hsa_circ_0085740, consistent with their established functions in stabilizing oncogenic RNAs. In contrast, METTL3 and DGCR8 interacted exclusively with downregulated circRNAs, suggesting that impaired processing or maturation of certain circRNAs may contribute to their reduced abundance in tumour samples. Additional RBPs, including C22ORF28, EWSR1, TAF15, and select LIN28 and FXR family members, preferentially bound upregulated circRNAs, whereas U2AF65 and ZC3H7B were enriched among interactions with downregulated species. This comprehensive circRNA-RBP interaction network reveals that dysregulated circRNAs in lung cancer exhibit non-random, expression-dependent RBP binding patterns, with upregulated circRNAs demonstrating enhanced multi-RBP associations that likely confer increased molecular stability and oncogenic functionality. The identification of EIF4A3, AGO2, and HuR as master regulatory hubs targeting both circRNA cohorts, alongside the selective engagement of IGF2BP family members with upregulated circRNAs and METTL3/DGCR8 with downregulated species, suggests that aberrant circRNA-RBP interactions constitute a fundamental mechanism driving lung cancer progression by rewiring post-transcriptional regulatory networks that control tumour cell proliferation, survival, metastasis, and therapeutic resistance. The final panel of 34 high-confidence circRNAs, supported by GO and KEGG enrichment results, converged on key pathways linked to transcriptional regulation, EMT, metabolic reprogramming, and invasive signalling. Collectively, these findings indicate that LUAD progression is accompanied by coordinated disruption of circRNA networks, characterized by loss of regulatory circRNAs and selective amplification of those promoting proliferation, metabolic flexibility, and tumour aggressiveness. The functional characterization of circRNA-miRNA-mRNA regulatory axes through GO and KEGG enrichment analyses offers critical insights into the molecular mechanisms underlying LUAD progression. By examining upregulated and downregulated gene sets independently, the analysis uncovers direction-specific biological alterations driven by circRNA dysregulation, thereby clarifying their distinct contributions to tumour behaviour. This integrative framework delineates the biological processes, cellular structures, and molecular functions most influenced by circRNA-mediated regulation, while parallel KEGG pathway enrichment broadens the understanding of how these genes converge on cancer-relevant signalling networks. Upregulated circRNA-associated genes demonstrated enrichment for functions related to elevated transcriptional activity, epithelial remodelling, vesicle trafficking, and cytoskeletal reorganization. Prominent GO terms such as DNA-templated transcription, epithelial proliferation, cell-cell adhesion, and neuron projection morphogenesis highlight enhanced transcriptional output, structural plasticity, and increased membrane dynamics. Correspondingly, enriched cellular components such as recycling endosomes, exocytic vesicle membranes, synaptic vesicles, and cortical ER suggest intensified intracellular transport and signalling turnover. Molecular function enrichment in transporter activity, kinase activation, PDZ-domain binding, and actin-binding reflects heightened metabolic and regulatory demands associated with tumour expansion and invasion. Together, these observations point toward upregulated circRNA networks promoting proliferative, invasive, and metabolically adaptive programs in LUAD. In contrast, downregulated circRNA-associated genes were linked to repression of transcriptional regulators, apoptotic mediators, adhesion structures, and signalling modulators. Key biological processes including attenuated gene expression control, reduced apoptotic regulation, and suppressed intracellular signalling indicating weakening of homeostatic and anti-tumour safeguards. The downregulation of focal adhesions, mitochondrial and ER membranes, and nuclear regulatory domains further highlights compromised structural integrity and organelle-associated regulation. Reduced DNA-binding transcription activator activity, kinase binding, and GTPase regulatory functions also suggest a loss of regulatory precision in signal transduction. The KEGG pathway analysis shows a strong functional difference between upregulated and downregulated circRNA-miRNA regulatory networks in LUAD. The pathways linked to upregulated circRNAs regulated metabolic activity, cellular stress adaptation, immune-responsive signalling, and cytoskeletal remodelling. These also included infection-related pathways, senescence, cancer-associated cascades, TGF-β signalling, glycolysis/gluconeogenesis, amino-acid metabolism, and regulation of the actin cytoskeleton, collectively reflecting a shift toward pathways that support tumour growth, plasticity, and survival. Conversely, downregulated circRNAs were associated with the suppression of key oncogenic and regulatory signalling systems. Attenuated pathways included cAMP, cGMP-PKG, MAPK, ErbB signalling, adherens junctions, axon guidance, and multiple cancer-linked pathways, along with endocrine regulatory modules such as aldosterone and growth hormone synthesis. The involvement of viral response and structural pathways further indicates a reduction in cellular coordination, adhesion, and signalling fidelity. Overall, these complementary patterns suggest that LUAD progression is facilitated by the simultaneous activation of pro-tumorigenic metabolic and signalling programs and the weakening of critical regulatory, structural, and homeostatic pathways highlighting the central role of dysregulated circRNA–miRNA networks in reshaping the LUAD molecular landscape. Conclusion In summary, this work identifies 34 consistently dysregulated circRNAs in LUAD through differential expression patterns, regulatory network mapping, and multi-layer validation. Our three-pronged analytical approach enhances biomarker identification in circRNA research through deep learning, clustering, and statistical methods. The consensus across all methods for candidates such as hsa_circ_0058622 (SP100) and hsa_circ_0085740 (PTK2) provides strong support that these signals are biologically meaningful rather than artifacts of sequencing noise. Furthermore, several high-ranking circRNAs such as hsa_circ_0043632 (circKRT17), hsa_circ_0085751 and hsa_circ_0085740 (circPTK2), hsa_circ_0024109 (circMMP1), and hsa_circ_0060931 (CYP24A1) have been previously reported in other cancer contexts, offering independent corroboration for their functional relevance. Functional enrichment analyses highlight their involvement in transcriptional regulation, epithelial remodelling, metabolic adaptation, and cancer-associated signalling pathways, while predicted miRNA and RBP interactions further illuminate their possible regulatory roles. These integrated insights nominate several circRNAs as strong biomarker and therapeutic candidates. Collectively, these circRNAs hold promise for early detection, prognosis, treatment stratification, prediction of therapeutic response, and even development of RNA-based interventions such as antisense oligonucleotides or circRNA mimics. However, since this study is entirely based on computational prediction model, further experimental validation is essential to confirm their expression patterns, molecular interactions, and functional relevance before advancing toward clinical or translational applications. Declarations Acknowledgement: The authors acknowledge the Bioinformatics Centre at Department of Bioscience and Biotechnology at Banasthali Vidyapith supported by DBT, Government of India, and the DST-FIST program of Government of India. Funding The authors received no financial support for the research, authorship, and/or publication of this article. Authors’ contributions: AM was responsible for data collection and curation, conducting the analyses, visualizing the results, and preparing and extensively revising the initial manuscript draft. RB contributed to the conceptualization of the study, supervised the research, provided computational resources, and participated in the critical review and editing of the manuscript. Conflict of Interests: The authors have declared that no conflict of interest exists. Data Availability Statement All datasets analyzed in this study are publicly available in the repositories and databases cited within the Methodology section. While the technical context is provided via accession numbers in the text, direct URLs to the datasets are provided below to facilitate immediate access https://www.ebi.ac.uk/ena/browser/view/SRP335547 https://www.ebi.ac.uk/ena/browser/view/SRP089923 https://www.ebi.ac.uk/ena/browser/view/SRP048484 https://www.ebi.ac.uk/ena/browser/view/SRP047399 https://www.circbase.org/ https://www.mirbase.org/ Ethics approval and consent to participate This study did not involve human participants or animals; therefore, ethical approval is not required. References Chaitanya Thandra K, Barsouk A, Saginala K, Aluru JS, Barsouk A. Epidemiology of lung cancer. Współczesna Onkologia. 2021;25(1):45–52. 10.5114/wo.2021.103829 . Sausville LN, Jones CC, Aldrich MC, Blot WJ, Pozzi A, Williams SM. Genetic variation in the eicosanoid pathway is associated with non-small-cell lung cancer (NSCLC) survival. PLoS ONE. Jul. 2017;12(7):e0180471. 10.1371/journal.pone.0180471 . Seguin L, Durandy M, Feral CC. Lung Adenocarcinoma Tumor Origin: A Guide for Personalized Medicine, Cancers (Basel) , vol. 14, no. 7, p. 1759, Mar. 2022, 10.3390/cancers14071759 Chen Z, Fillmore CM, Hammerman PS, Kim CF, Wong K-K. Non-small-cell lung cancers: a heterogeneous set of diseases. Nat Rev Cancer. Aug. 2014;14(8):535–46. 10.1038/nrc3775 . Chen L-L. The biogenesis and emerging roles of circular RNAs. Nat Rev Mol Cell Biol. Apr. 2016;17(4):205–11. 10.1038/nrm.2015.32 . Cocquerelle C, Mascrez B, Hétuin D, Bailleul B. Mis-splicing yields circular RNA molecules. FASEB J. Jan. 1993;7(1):155–60. 10.1096/fasebj.7.1.7678559 . Jeck WR, et al. Circular RNAs are abundant, conserved, and associated with ALU repeats. RNA. Feb. 2013;19(2):141–57. 10.1261/rna.035667.112 . Salzman J, Chen RE, Olsen MN, Wang PL, Brown PO. Cell-Type Specific Features of Circular RNA Expression, PLoS Genet , vol. 9, no. 9, p. e1003777, Sep. 2013, 10.1371/journal.pgen.1003777 Malviya A, Bhuyan R. The recent advancements in circRNA research: From biogenesis to therapeutic interventions. Pathol Res Pract. Aug. 2023;248:154697. 10.1016/j.prp.2023.154697 . Leinonen R, Sugawara H, Shumway M. The Sequence Read Archive. Nucleic Acids Res. Jan. 2011;39:D19–21. 10.1093/nar/gkq1019 . Database. Harrison PW, et al. The European Nucleotide Archive in 2020. Nucleic Acids Res. Jan. 2021;49:D82–5. 10.1093/nar/gkaa1028 . Kent WJ et al. Jun., The Human Genome Browser at UCSC, Genome Res , vol. 12, no. 6, pp. 996–1006, 2002, 10.1101/gr.229102 Glažar P, Papavasileiou P, Rajewsky N. CircBase: A database for circular RNAs, RNA , vol. 20, no. 11, pp. 1666–1670, Nov. 2014, 10.1261/rna.043687.113 Kozomara A, Birgaoanu M, Griffiths-Jones S. miRBase: from microRNA sequences to function. Nucleic Acids Res. Jan. 2019;47:D155–62. 10.1093/nar/gky1141 . no. D1. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. Aug. 2014;30(15):2114–20. 10.1093/bioinformatics/btu170 . Gao Y, Wang J, Zhao F. CIRI: an efficient and unbiased algorithm for de novo circular RNA identification. Genome Biol. Jan. 2015;16(1). 10.1186/s13059-014-0571–3 . Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics. Jul. 2009;25(14):1754–60. 10.1093/bioinformatics/btp324 . Pertea M, Pertea GM, Antonescu CM, Chang T-C, Mendell JT, Salzberg SL. StringTie enables improved reconstruction of a transcriptome from RNA-seq reads, Nat Biotechnol , vol. 33, no. 3, pp. 290–295, Mar. 2015, 10.1038/nbt.3122 Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. Dec. 2014;15:550. 10.1186/s13059-014-0550–8 . Miranda KC et al. Sep., A Pattern-Based Method for the Identification of MicroRNA Binding Sites and Their Corresponding Heteroduplexes, Cell , vol. 126, no. 6, pp. 1203–1217, 2006, 10.1016/j.cell.2006.07.031 Shannon P et al. Nov., Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks, Genome Res , vol. 13, no. 11, pp. 2498–2504, 2003, 10.1101/gr.1239303 Dudekula DB, Panda AC, Grammatikakis I, De S, Abdelmohsen K, Gorospe M. CircInteractome: A web tool for exploring circular RNAs and their interacting proteins and microRNAs. RNA Biol. Jan. 2016;13(1):34–42. 10.1080/15476286.2015.1128065 . Ru Y, et al. The multiMiR R package and database: integration of microRNA–target interactions along with their disease and drug associations. Nucleic Acids Res. Sep. 2014;42:e133–133. 10.1093/nar/gku631 . Riffo-Campos Á, Riquelme I, Brebi-Mieville P. Tools for Sequence-Based miRNA Target Prediction: What to Choose? Int J Mol Sci , vol. 17, no. 12, p. 1987, Dec. 2016. 10.3390/ijms17121987 Chen EY, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. Dec. 2013;14(1):128. 10.1186/1471-2105-14–128 . Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. Jan. 2017;45:D353–61. 10.1093/nar/gkw1092 . Malviya A, Bhuyan R. Circular RNAs in cancer: roles, mechanisms, and therapeutic potential across colorectal, gastric, liver, and lung carcinomas. Discover Oncol. Jan. 2025;16(1). 10.1007/s12672-025-01743–9 . Wang L, et al. Circular RNA hsa_circ_0008305 (circPTK2) inhibits TGF-β-induced epithelial-mesenchymal transition and metastasis by controlling TIF1γ in non-small cell lung cancer. Mol Cancer. Sep. 2018;17(1):140. 10.1186/s12943-018-0889–7 . Yang H, et al. CircPTK2 (hsa_circ_0005273) as a novel therapeutic target for metastatic colorectal cancer. Mol Cancer. Jan. 2020;19(1):13. 10.1186/s12943-020-1139–3 . Dai W et al. Aug., CircMMP1 promotes colorectal cancer growth and metastasis by sponging miR–1238 and upregulating MMP family expression, Ann Transl Med , vol. 9, no. 16, pp. 1341–1341, 2021, 10.21037/atm–21 –3930. Ji Y, Zhao Q, Feng W, Peng Y, Hu B, Chen Q. N6-Methyladenosine Modification of CIRCKRT17 Initiated by METTL3 Promotes Osimertinib Resistance of Lung Adenocarcinoma by EIF4A3 to Enhance YAP1 Stability, Cancers (Basel) , vol. 14, no. 22, p. 5582, Nov. 2022, 10.3390/cancers14225582 Wu X et al. Feb., CircCYP24A1 hampered malignant phenotype of renal cancer carcinoma through modulating CMTM–4 expression via sponging miR–421, Cell Death Dis , vol. 13, no. 2, p. 190, 2022, 10.1038/s41419-022-04623–0 Li R, Chai L, Lei L, Guo R, Wen X. MiR–671–5p sponging activity of circMMP1 promotes esophageal squamous cancer progression, Thorac Cancer , vol. 14, no. 29, pp. 2924–2933, Oct. 2023, 10.1111/1759–7714.15078 . Zou Q, Zhang Y, Zhu D, Liu X, Wang C, Xiang H. CircMMP11 as a prognostic biomarker mediates miR–361–3p/HMGB1 axis to accelerate malignant progression of hepatocellular carcinoma. Open Med. Nov. 2023;18(1). 10.1515/med–2023–0803 . Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. Jan. 2022;12(1):31–46. 10.1158/2159–8290.CD–21 . Roizen MF. Hallmarks of Cancer: The Next Generation, Yearbook of Anesthesiology and Pain Management , vol. 2012, p. 13, Jan. 2012. 10.1016/j.yane.2012.02.046 Additional Declarations No competing interests reported. Supplementary Files suptab.docx supfig.docx HL.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-8684357","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":593503035,"identity":"5bfb9a0d-3320-4de8-89c2-66288fa26417","order_by":0,"name":"Ayushi Malviya","email":"","orcid":"","institution":"Banasthali University","correspondingAuthor":false,"prefix":"","firstName":"Ayushi","middleName":"","lastName":"Malviya","suffix":""},{"id":593503036,"identity":"d5be6b0c-8bae-475f-8498-9c60bd15c6fd","order_by":1,"name":"Rajabrata Bhuyan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+UlEQVRIiWNgGAWjYHCCBGYwxczYcOADkGZjJ1KLBAN788GDM0BamImwBqKF51jyYR4YFx8wbz/w8HPhHrs6/hk5Bodtfm2T52NmYPzwMQe3FpkzCcnSM54lS0jcAGrJ7btt2MbMwCw5cxtuLRIMCQnSPAeYJRjAWnpuMwK1sDHz4tPC/yD5N8+Begl5kBbLntv2hLVIJKQBbTksYXDmWMJhhh+3E4nQ8iDNesaB45IbjzcfONjbcDu5jZmxGb9f+HOSbxccqOaXO8zY/OHHn9u289ubD374iEcLAwNPAoLN2AYmG/CpBwL2A0icPwQUj4JRMApGwYgEAKKLVJ/NNwnlAAAAAElFTkSuQmCC","orcid":"","institution":"Banasthali University","correspondingAuthor":true,"prefix":"","firstName":"Rajabrata","middleName":"","lastName":"Bhuyan","suffix":""}],"badges":[],"createdAt":"2026-01-24 06:39:49","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8684357/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8684357/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102990836,"identity":"642c0b98-ef0d-48c8-9fed-97ac82a1aca9","added_by":"auto","created_at":"2026-02-19 11:27:04","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":733637,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eClassification and characterization of detected circRNAs.\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e (a) \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eThe donut plot displays the distribution of detected circRNAs classified as either Known or Novel based on their prior documentation in the Circbase database; \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(b) A pie chart illustrating the genomic annotation of the known circRNAs with the majority being exonic, with other categories including overlapping, intronic, non-coding, and antisense circRNAs, based on their genomic origin or gene context\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/552db5838403fb8b434087a9.png"},{"id":102990794,"identity":"526d21b3-230c-46f9-a694-0c04710630d8","added_by":"auto","created_at":"2026-02-19 11:26:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1282944,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eDifferential expression of circRNAs. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(a)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Heatmap displaying normalized circRNA expression (Z-scores) of the top 34 prioritized circRNAs. Rows represent selected circRNAs, and columns denote individual samples. Expression patterns reveal clustering among circRNAs and samples based on expression similarity; \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(b)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Volcano plot depicting differential circRNA expression in LUAD samples. The plot displays log\u003c/em\u003e\u003csub\u003e\u003cem\u003e2\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e fold changes versus -log\u003c/em\u003e\u003csub\u003e\u003cem\u003e10\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e adjusted p-values for detected circRNAs, highlighting 625 upregulated and 855 downregulated circRNAs. Circles represent individual circRNAs classified as upregulated, downregulated, or not significant based on statistical thresholds; \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(c)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Principal component analysis (PCA) plot showing separation between LUAD Cancer and normal samples based on circRNA expression. PC1 and PC2 together explain major variance, with distinct clustering observed for each group\u003c/em\u003e.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/cffeb94e1db0b0d3c5ada834.png"},{"id":102990834,"identity":"e560c0f5-79f9-4b68-b6fd-eb839bff93d1","added_by":"auto","created_at":"2026-02-19 11:27:03","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1270789,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eNetwork visualization of highly dysregulated circRNAs and their predicted interacting miRNAs in LUAD. Rectangular Nodes represent individual circRNAs and Diamond shaped nodes represent miRNAs, and edges show predicted regulatory relationships, highlighting possible regulatory roles in LUAD progression. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(a)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Depiction of upregulated circRNAs (red, rectangular nodes) and their miRNA associations (teal, diamond shaped nodes) and \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(b)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003eDownregulated circRNAs (green, rectangular nodes) and their miRNA (teal, diamond shaped nodes) associations.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/4725feb2ec2513d707a0119a.png"},{"id":102990923,"identity":"ec03cda9-6f6f-4de7-a857-bc1c0e35b506","added_by":"auto","created_at":"2026-02-19 11:27:10","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2630801,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eNetwork visualization of circRNA interactions with RNA-binding proteins (RBPs) in LUAD samples. Nodes represent circRNAs (green: downregulated, red: upregulated) and RBPs (purple: targeted by both circRNA categories, light green: targeted only by downregulated circRNAs, yellow: targeted only by upregulated circRNAs). Edges indicate predicted binding interactions, with colour coordination reflecting circRNA regulation status (green edges for downregulated circRNAs, red edges for upregulated circRNAs). This network highlights the complex regulatory relationships between circRNAs and RBPs involved in LUAD.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/00c82510120998fb7d8b1032.png"},{"id":102990762,"identity":"62b5b259-07e0-493f-9f3f-b7be542bc89d","added_by":"auto","created_at":"2026-02-19 11:26:45","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1304603,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eOverview of circRNA prioritization results across multiple analytical methods. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(a)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Consensus ranking of circRNAs across analytical methods with the plot depicting top circRNAs with similar consensus rankings derived from MLP, cluster, and statistical (stats) algorithms, indicating strong agreement among methods in identifying consistently significant circRNA candidates;\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e (b)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Venn diagram illustrating the overlapping top circRNAs identified via all three algorithms; \u003c/em\u003e\u003cem\u003e\u003cstrong\u003e(c) \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eDensity plots showing the distribution of circRNA scores obtained from each algorithm, illustrating comparable scoring trends and overlapping regions that reflect methodological consistency in circRNA prioritization.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/51020604ef2c7713f4ad236f.png"},{"id":102990833,"identity":"b62bf3cb-b91a-41c6-9672-94496aaf95ce","added_by":"auto","created_at":"2026-02-19 11:27:03","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":495675,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePathway enrichment of target genes of dysregulated circRNAs.\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e (a)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Bubble plot illustrating the enriched pathways targeted by Upregulated circRNAs;\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e (b)\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Bubble plot illustrating the enriched pathways targeted by Downregulated circRNAs; the plot provides insights into the potential signalling pathways influenced by dysregulated circRNAs in each case, contributing to our understanding of their roles in lung cancer development and progression.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/60c8023e8ad59aed78e41537.png"},{"id":108491072,"identity":"4f35bd2b-6250-4436-bba4-5845d9cdea8a","added_by":"auto","created_at":"2026-05-05 09:51:57","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8555788,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/cd538197-50d5-473a-b105-44c4d9e3fc49.pdf"},{"id":102990974,"identity":"343132ed-6016-4722-9444-c39e8a869c48","added_by":"auto","created_at":"2026-02-19 11:27:25","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":616385,"visible":true,"origin":"","legend":"","description":"","filename":"suptab.docx","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/e43dd95c86fd8a2c614f51d1.docx"},{"id":102990930,"identity":"8d4aaaf2-a7ca-442b-ac5b-ebc45adb1801","added_by":"auto","created_at":"2026-02-19 11:27:12","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":775377,"visible":true,"origin":"","legend":"","description":"","filename":"supfig.docx","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/b41b00324d228d9f37416bad.docx"},{"id":102990918,"identity":"52757646-c54f-4eeb-8539-05f6c7c77aaf","added_by":"auto","created_at":"2026-02-19 11:27:09","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":5454,"visible":true,"origin":"","legend":"","description":"","filename":"HL.docx","url":"https://assets-eu.researchsquare.com/files/rs-8684357/v1/9b1549e049961761817dfc6e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrated transcriptomic and machine learning-driven analysis reveals high-confidence circular RNA biomarkers in Lung Adenocarcinoma","fulltext":[{"header":"Introduction","content":"\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eLung cancer is responsible for over 1.8\u0026nbsp;million deaths each year, continues to pose a significant global health risk, and is recognised as the primary cause of cancer-related fatalities worldwide\u003c/span\u003e [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Its five-year survival rate \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eremains extremely low despite developments in diagnosis and treatment, especially for non-small cell lung cancer (NSCLC), which accounts for approximately 85% of cases\u003c/span\u003e [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Lung adenocarcinoma, the most common subtype of NSCLC, has a consistently poor prognosis, and recurrent resistance to targeted therapy [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. The molecular mechanisms underlying lung cancer involve complex regulatory networks including genetic mutations, epigenetic alterations, and dysregulated non-coding RNAs, among which circRNAs have \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eemerged as critical modulators of tumour initiation, progression, and drug resistance\u003c/span\u003e [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Circular RNAs (circRNAs) are generated through alternative splicing events that produce covalently closed loop structures, conferring remarkable resistance to exonuclease-mediated degradation, and thereby enhancing their stability [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. These molecules \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003efrequently display tissue-specific and developmentally regulated expression patterns\u003c/span\u003e [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Initially viewed as transcriptional artifacts, they are now recognized as multifunctional regulators that modulate gene expression by sponging microRNAs, binding RNA-binding proteins (RBPs), and in some cases, \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003efacilitating cap-independent translation of bioactive peptides. This evolving understanding has positioned circRNAs as promising biomarkers and potential therapeutic targets in various malignancies\u003c/span\u003e [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eWhile several circRNAs have been implicated in lung cancer progression through modulation of proliferation, invasion, metastasis, and drug resistance, our understanding of their full functional spectrum remains incomplete. This study presents a comprehensive exploration of the functional landscape of circRNAs in lung cancer, integrating multi-level analyses of circRNA-miRNA-mRNA regulatory networks, and RBP interactions. By leveraging state-of-the-art AI/ML approaches, we prioritized statistically significant and biologically functional circRNA candidates and predicted their biomarker potential. The aim of the study is to elucidate the new regulatory roles of circRNAs establishing a basis for the next generation of circRNA-based diagnostics and therapies, offering insights poised to transform the lung cancer research.\u003c/span\u003e \u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eDataset collection\u003c/span\u003e\u003c/h2\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe NCBI Sequence Read Archive (SRA) was systematically explored to retrieve publicly available RNA-Seq datasets optimized for circRNA detection\u003c/span\u003e [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eOnly transcriptomic paired end reads from LUAD samples enriched for non-coding RNAs were considered. The selected datasets were categorized based on their intended application: ncRNA-enriched libraries were retained for circRNA identification (Accession IDs: SRP335547; SRP089923; SRP048484; SRP047399), while libraries containing both cancerous and non-cancerous samples were utilized for differential expression analyses. All RNA-Seq libraries were obtained from the European Nucleotide Archive\u003c/span\u003e [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe human reference genome (GRCh37.fa) along with its corresponding annotation file (ucsc_GRCh37.gtf) was downloaded from the UCSC Genome Browser\u003c/span\u003e [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eFor annotation of identified circRNAs, reference files from the CircBase database (hsa_hg19_circRNA.bed; hsa_hg19_circRNA.gtf) were employed\u003c/span\u003e [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. In addition, mature and annotated miRNA sequences were retrieved from the \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003emiRBase database\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003ehomo_sapiens.fa)\u003c/span\u003e [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCircRNA detection\u003c/h3\u003e\n\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003ePre-processing and quality control of each dataset were conducted using FastQC {\u003c/span\u003e \u003cspan class=\"ExternalRef\"\u003e \u003cspan class=\"RefSource\"\u003ehttp://www.bioinformatics.babraham.ac.uk/projects/fastqc/\u003c/span\u003e \u003cspan address=\"http://www.bioinformatics.babraham.ac.uk/projects/fastqc/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e \u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003e} to assess read quality. Reads with a base quality score below Q20 were dropped and the sequences were trimmed using Trimmomatic\u003c/span\u003e [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. The quality-filtered reads were then subjected to circRNA detection using CIRI2, the updated version of the CIRI (circRNA identifier) algorithm, which relies on chiastic clipping signals and multiple filtration steps to reliably and unbiasedly detect circRNAs [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eIt utilises BWA for spliced alignment against the human reference genome\u003c/span\u003e [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eCIRI2 was chosen for this study because it offers a strong balance of accuracy, computational efficiency, and low resource demand. CircRNA detection was performed individually for each sample, and the resulting output files were subsequently merged into a consolidated BED file for downstream analyses.\u003c/span\u003e\u003c/p\u003e\n\u003ch3\u003eDifferential gene expression studies\u003c/h3\u003e\n\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eStringTie\u003c/span\u003e [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] was used for transcript normalization, abundance estimation, and generation of count data for downstream analysis using the Bioconductor package DESeq2 [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Differential expression analysis was performed using TPM-normalized count values. CircRNA transcripts from cancer and normal samples were further filtered based on an adjusted p-value threshold of 0.05, followed by refinement using a log\u003csub\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003e2\u003c/span\u003e\u003c/sub\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003efold change cut-off of \u0026plusmn;\u0026thinsp;2.\u003c/span\u003e\u003c/p\u003e\n\u003ch3\u003ePrediction of target miRNAs and RBPs\u003c/h3\u003e\n\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe interactions between significantly differentially expressed circRNAs and their target miRNAs were predicted using the miRanda tool\u003c/span\u003e [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], applying a threshold score of 150 and a minimum free energy cutoff of \u0026minus;\u0026thinsp;25 to ensure the inclusion of a sufficiently comprehensive pool for feature extraction \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003erequired in the next steps. An interaction network comprising the top-ranking circRNAs and their high-confidence miRNA targets was subsequently constructed and visualized using Cytoscape\u003c/span\u003e [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eSimultaneously, target RNA-binding proteins (RBPs) of the significantly dysregulated circRNAs were retrieved from the circInteractome database\u003c/span\u003e [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], and circRNA-RBP interaction networks were generated in \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eCytoscape for both upregulated and downregulated circRNAs.\u003c/span\u003e\u003c/p\u003e\n\u003ch3\u003eMachine Learning-mediated biomarker potential prediction\u003c/h3\u003e\n\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eA circRNA feature table was constructed containing six biologically relevant attributes: Exclusivity Score, Consistency Score, Differential Expression Score (DE_Score), Binding Affinity, Regulatory Potential, and Sponge Capacity Score (\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003eSupplementary Table\u0026nbsp;1)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThis multi-dimensional dataset provides a comprehensive representation of circRNA functionality and biomarker potential. Three complementary computational frameworks were applied on this feature table. First, a feature-weighted statistical scoring approach generated a Biomarker Potential Score through the integration of weighted features, yielding an interpretable baseline ranking. Second, unsupervised machine learning (PCA, K-means clustering, and Isolation Forest) was used to identify intrinsic patterns, stratify circRNAs into high/medium/low groups, and highlight potential outliers. A normalized, weighted biomarker score was calculated within these clusters. Third, a deep learning model was implemented using a multilayer neural network, operating in supervised or semi-supervised mode, to capture nonlinear feature interactions and assign probabilistic scores of biomarker relevance. Consensus circRNAs were defined as those consistently prioritized by all three methods. This integrative, multi-tiered strategy reduced false positives, enhanced reproducibility, and outperformed traditional single-parameter screening by ensuring a more robust and biologically meaningful selection of circRNA biomarker candidates.\u003c/span\u003e\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eGene Ontology and pathway enrichment studies\u003c/span\u003e\u003c/h2\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eTo identify the miRNA-targeted genes for subsequent GO analysis, the multimiR\u003c/span\u003e [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] package in R was employed. Three target prediction databases were queried (miRAnda, \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003emirbase and TargetScan)\u003c/span\u003e [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], and only those genes that were predicted by at least two of the three databases and/or were experimentally \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003evalidated were included in the final list. Functional annotation of significantly dysregulated circRNA target genes, identified through their interacting miRNAs, was performed using Gene Ontology (GO) analysis via the Enrichr server\u003c/span\u003e [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Gene functions were categorized into three domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). These categories were separately \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eanalyzed for the upregulated and downregulated gene sets. Additionally, pathway enrichment analysis was conducted using the KEGG database\u003c/span\u003e [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] .\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003eCharacterization of circRNA expression and host gene associations\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe circRNA profile in LUAD was detected using the CIRIv2 tool and subsequently annotated with circBase database. CIRI2 employs maximum likelihood-based multiple seed matching to effectively distinguish true back-splice junction (BSJ) reads from incorrectly mapped non-BSJ reads, enabling both low false-discovery rates and high sensitivity. Its enhanced improved read-classification accuracy, combined with multithreading support, allows for faster execution and more efficient RAM usage. This systematic and rigorous workflow facilitated the accurate identification, distinction, and classification of circRNAs using the circbase database, allowing them to be organized into two primary categories: previously unreported\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003enovel circRNAs\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eand those already catalogued as\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003eknown circRNAs (\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003e\u0026ldquo;Novel circRNAs\u0026rdquo; are defined as circular RNAs identified in LUAD samples that do not overlap with, or correspond to, any circRNAs previously annotated in the circBase database, and thus represent newly identified and uncharacterized circRNA transcripts. Conversely, \u0026ldquo;known circRNAs\u0026rdquo; are those whose genomic coordinates and features match entries previously documented in circBase, representing validated or annotated circular RNA species.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eIn the LUAD dataset, 18,922 circRNAs were classified as Novel, while a larger number, 33,822 circRNAs, were identified as Known. The majority of circRNAs, both novel and known, mapped predominantly to exonic regions (32,452), consistent with current understanding that exonic circRNAs constitute the bulk of circRNA populations in cancer\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eA smaller number of circRNAs derived from intronic regions (1,194), reflecting the diversity of circRNA biogenesis beyond exons. Additionally, 2,656 circRNAs were categorized as overlapping, spanning multiple genomic features and indicating complex regulatory loci that may be hotbeds for alternative circularization or transcript overlap. Non-coding circRNAs totalled 1,199, highlighting the importance of circRNAs beyond protein-coding potential in regulatory networks. A minor subset of 86 antisense circRNAs was identified, indicating specific regulatory roles potentially linked to gene expression interference or silencing.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eDifferential expression of circRNAs in LUAD vs normal lung cells\u003c/h3\u003e\n\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eDifferential gene expression analysis revealed substantial transcriptomic alterations in lung adenocarcinoma compared to normal lung tissue. A total of 1,480 circRNAs exhibited significant differential expression (adjusted p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05), with 855 circRNAs demonstrating downregulation and 625 circRNAs showing upregulation in tumour samples\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThis pattern, characterized by a greater proportion of downregulation (57.8%) relative to their higher expression (42.2%), suggests that lung adenocarcinoma pathogenesis involves extensive transcriptional repression alongside oncogenic activation. Complementary insights into the expression patterns and sample clustering for circRNAs in LUAD study are offered by the accompanying heatmap and PCA plots\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eWhile the PCA plot, which is based on principal component analysis of circRNA expression profiles, visually distinguishes between cancer and normal samples, the heatmap shows the relative expression levels of certain circRNAs across different samples.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe predominance of downregulated circRNAs may reflect the loss of normal lung epithelial cell identity and differentiation programs during malignant transformation. Conversely, the upregulated circRNAs likely encompass oncogenic drivers, genes involved in proliferation, metabolic reprogramming, and immune evasion mechanisms characteristic of lung cancer. The bilateral nature of these expression changes, with substantial alterations in both directions, highlights the complex rewiring of cellular transcriptional networks that accompanies tumorigenesis and supports the multifactorial nature of lung adenocarcinoma development.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eCircRNA-miRNA-mRNA regulatory network and RBP interactions\u003c/span\u003e\u003c/h2\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eAnalysis of the upregulated circRNAs revealed a distinct interaction network\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003echaracterized by a smaller but relatively focused set of circRNAs that engage with specific miRNAs. Prominent circRNAs included hsa_circ_0012003, hsa_circ_0026425, hsa_circ_0027924, hsa_circ_0035015, hsa_circ_0043614, hsa_circ_0043632, hsa_circ_0044130, hsa_circ_0052195, hsa_circ_0085288, and hsa_circ_0085740. Several of these circRNAs acted as hub molecules by interacting with multiple miRNAs; for example, hsa_circ_0035015 (targeting hsa-miR-665, hsa-miR-770-5p, hsa-miR-4739), hsa_circ_0043614 (targeting hsa-miR-7158-5p, hsa-miR-6894-5p), and hsa_circ_0043632 (targeting hsa-miR-4776-3p, hsa-miR-7106-5p, hsa-miR-1293, hsa-miR-6791-3p). This pattern suggests that a few highly expressed circRNAs may exert substantial influence over a diverse repertoire of miRNAs.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eNetwork analysis of the\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003edownregulated circRNAs\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003erevealed a complex regulatory catalogue comprising several circRNAs, notably hsa_circ_0049036, hsa_circ_0046727, hsa_circ_0024652, hsa_circ_0092361, hsa_circ_0002734, hsa_circ_0012248, hsa_circ_0084884, hsa_circ_0000245, hsa_circ_0011819, hsa_circ_0082722, and hsa_circ_0079823, that interact with a wide spectrum of miRNAs, including hsa-miR-4632-5p, hsa-miR-330-5p, hsa-miR-3714, hsa-miR-519a-2-5p, hsa-miR-6088, hsa-miR-615-5p,hsa-miR-615-5p, hsa-miR-197-5p, hsa-miR-612, hsa-miR-15a-3p, and hsa-miR-324-3p, among others. Several circRNAs exhibited hub-like properties by targeting multiple miRNAs, suggesting that they may exert broad regulatory influence. Conversely, certain miRNAs such as hsa-miR-197-5p, hsa-miR-1285-3p, hsa-miR-619-5p, hsa-miR-363-5p, hsa-miR-7110-5p, and hsa-miR-328-5p were recurrently targeted by multiple circRNAs, indicating their central role in the post-transcriptional network.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe circRNA-RBP interaction network\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(\u003c/span\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e\u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003ein lung cancer reveals a complex regulatory architecture involving 27 differentially expressed circRNAs (13 upregulated, 14 downregulated) and 24 RNA-binding proteins exhibiting distinct interaction patterns. Network topology analysis demonstrates that upregulated circRNAs display significantly higher RBP connectivity compared to their downregulated counterparts, with hsa_circ_0005352 emerging as a central hub interacting with 17 RBPs, followed by hsa_circ_0085740 (10 RBPs) and hsa_circ_0044130 (6 RBPs). Among the RBPs identified, EIF4A3 functions as the most promiscuous binding partner, interacting with 22 circRNAs from both expression cohorts, while AGO2 targets 15 circRNAs and HuR binds 11 circRNAs, establishing these proteins as convergent regulatory hubs that bridge upregulated and downregulated circRNA populations. The IGF2BP protein family (IGF2BP1/2/3) demonstrates preferential binding to upregulated circRNAs, particularly hsa_circ_0005352, hsa_circ_0044130, and hsa_circ_0085740, suggesting their role in stabilizing oncogenic circRNA transcripts.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eNotably, METTL3 and DGCR8 exclusively interact with downregulated circRNAs, while C22ORF28, EWSR1, TAF15, and specific members of the LIN28 and FXR families show selective binding to upregulated circRNAs. Splicing factors U2AF65 and ZC3H7B uniquely associate with downregulated circRNAs, whereas stress granule components CAPRIN1 and FXR1/2 exhibit differential targeting patterns. The extensive interconnectivity of upregulated circRNAs with multiple RBPs suggests enhanced post-transcriptional stability mechanisms driving their accumulation in lung cancer cells\u003c/span\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eMachine Learning driven biomarker prediction\u003c/span\u003e\u003c/h2\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eWe identified 34 circular RNA (circRNA) candidates as high-potential biomarkers through a comprehensive multi-method consensus ranking approach that integrated three independent analytical frameworks: deep learning-based analysis (DL score), clustering-based assessment (Cluster score), and statistical calculation (Stats score). All candidates achieved DL scores exceeding 0.987, demonstrating robust discriminatory potential and were derived from genes with diverse biological functions. Each approach highlights different aspects of biological significance: the statistical model emphasizes disease specificity, consistency, and regulatory or sponge capacity; the unsupervised methods reveal circRNAs central to meaningful biological clusters or with high-impact outlier profiles; and the deep learning model captures nonlinear feature interactions predictive of disease relevance\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003e(Supplementary Table\u0026nbsp;1)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eWhile reducing false positives and boosting confidence in their possible therapeutic efficacy, the convergence of these orthogonal approaches on the same set of circRNAs highlights their multifaceted strengths, such as being structurally pivotal in regulatory networks, reproducible across datasets, and biologically distinctive.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eAmong upregulated candidates, hsa_circ_0085740 from PTK2 (focal adhesion kinase, LFC: 10.6) demonstrated exceptional performance with top 4 rankings in clustering and statistical analyses. The hsa_circ_0060931 from CYP24A1, a key enzyme in vitamin D metabolism, exhibited the highest fold-change, while hsa_circ_0052195 from PTPRH (protein tyrosine phosphatase receptor, LFC: 8.97) maintained consistent top 10 rankings across all methods. Other key upregulated candidates included hsa_circ_0058736 from C2orf82, hsa_circ_0024109 from MMP1 (matrix metalloproteinase involved in extracellular matrix remodelling), and hsa_circ_0043632 from KRT17 (keratin 17, a cytoskeletal protein).The top-ranked downregulated circRNAs included hsa_circ_0058622 from the SP100 gene, which exhibited exceptional consensus with rank 1 across all three methods, followed by hsa_circ_0012248 from NASP gene and hsa_circ_0037308 from TELO2 gene, both showing consistent top-5 rankings. Notably, hsa_circ_0016392 derived from PPP2R5A, a regulatory subunit of protein phosphatase 2A, displayed the most extreme downregulation (LFC: -13.21) with strong statistical support. Additional downregulated candidates of interest included hsa_circ_0048053 from SHC2 (adaptor protein in signalling pathways), hsa_circ_0046727 from SMCHD1 (chromatin modifier), and hsa_circ_0003663 from TOP3A (DNA topoisomerase), all demonstrating coordinated top 10 rankings.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe top 34 dysregulated circRNAs (17 upregulated and 17 downregulated) (\u003c/span\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cspan type=\"BoldItalicSmallCaps\" class=\"BoldItalicSmallCaps\" name=\"Emphasis\"\u003e) were\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003esubsequently selected as the final high-confidence biomarker panel for downstream functional validation, including Gene Ontology (GO) term analyses and pathway enrichment studies, to investigate their roles in lung cancer biology. The enrichment results provided independent evidence linking these circRNAs to processes such as transcriptional regulation, EMT, metabolic adaptation, and invasive signalling, further reinforcing their relevance to lung cancer progression. This integrative workflow demonstrates that multi-algorithm consensus combined with biological pathway validation strengthens biomarker discovery.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eList of top-scoring circRNAs identified across three independent ranking algorithms\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecircRNA\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGene name\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eExpression\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLFC value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eDL score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eDL rank\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eCluster score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eCluster rank\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eStats score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eStats rank\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0058622\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSP100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-12.17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99663204\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.757\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.740\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0012248\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNASP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-11.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99612737\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.744\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.727\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0037308\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTELO2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-9.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9952877\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.728\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0000245\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNone\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-2.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9949515\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.608\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.650\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0085751\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePTK2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99449205\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.558\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.643\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0048053\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSHC2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-9.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99448025\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.716\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.700\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0046727\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSMCHD1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-9.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99440217\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.713\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.696\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0052195\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePTPRH\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9940776\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.707\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.695\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0058736\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2orf82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99389577\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.587\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.671\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0085740\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePTK2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99388254\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.716\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0084884\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTMEM67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-5.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99361026\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.572\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.655\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0003663\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTOP3A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-8.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9934958\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.696\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.679\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0092334\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePRELID2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9927882\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.537\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.630\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e23\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0005352\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCDCA2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9924915\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.539\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.629\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e24\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0024109\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMMP1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99242914\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.575\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.659\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0011819\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMACF1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-3.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9921859\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.636\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.631\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0024652\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGRIK4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9918901\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.538\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.624\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e26\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0060931\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCYP24A1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e28.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9917231\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.579\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.645\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0085288\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCTHRC1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9914436\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.538\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.636\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0043632\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKRT17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99133587\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.553\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.643\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0092004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFLNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-3.98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99129635\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.517\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.612\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0016392\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePPP2R5A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-13.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.991119\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.690\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.639\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0092361\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVIPR1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-4.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99085486\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.545\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.635\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0026425\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKRT6A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99053043\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.548\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.634\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0079823\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBMPER\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-4.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9904053\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.553\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.647\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0079251\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSLC29A4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9901022\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.503\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.602\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0044130\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKIF18B\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9897237\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.508\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.606\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0002734\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSEMA5A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-2.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.98954356\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.669\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.627\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0035015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCKMT1B\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9894872\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.525\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.618\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0027924\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC12orf48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.66\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.98910236\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.499\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.599\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e49\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0049036\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFBN3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-3.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.98865277\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.5117\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.605\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e34\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0012003\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSLC2A1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9880233\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.497\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.603\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e37\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0043614\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKRT14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.98796755\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.526\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.619\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehsa_circ_0082722\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDENND2A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-2.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.9872055\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.501\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.598\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e46\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"10\"\u003e\u003cb\u003eDL score\u003c/b\u003e: Deep learning-based score; \u003cb\u003eCluster score\u003c/b\u003e: Clustering based score; \u003cb\u003eStats scores\u003c/b\u003e: Statistically calculated scores; All three represent independent ranking methods; LFC shows expression fold change.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eTarget gene set enrichment analysis\u003c/span\u003e\u003c/h2\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe target genes of circRNA-miRNA regulatory axis were subjected to Gene Ontology (GO) studies aimed at characterizing the biological processes underlying mechanism of Lung Cancer progression, to provide valuable insights into their pathogenesis. The Gene functions were systematically categorized into the three major Gene Ontology (GO) domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). To capture directional biological changes, GO enrichment analyses were performed separately for the upregulated and downregulated gene sets, enabling the identification of distinct functional regulation associated with each expression pattern. In parallel, pathway enrichment analysis was conducted using the KEGG database. This integrated GO-KEGG framework enabled a comprehensive investigation of the functional landscape, facilitating the identification of key pathways, cellular modules, and molecular mechanisms potentially driving disease progression\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe GO enrichment analysis of genes associated with upregulated circRNAs revealed strong activation of pathways linked to transcriptional activity, epithelial remodelling, and membrane-associated processes (\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003eSupplementary Fig.\u0026nbsp;1\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003e). An elevation in overall transcriptional activity was observed through key BP terms enrichment, including DNA-templated transcription, RNA biosynthetic processes, and RNA polymerase II\u0026ndash;mediated transcription. Enrichment of epithelium development, positive regulation of epithelial cell proliferation, and cell-cell adhesion, with enhanced proliferative and structural dynamics within epithelial cells were observed. Additionally, prominent terms such as neuron projection development and projection morphogenesis, suggesting cytoskeletal and structural reorganization, and cellular plasticity were inferred. The CC terms associated with upregulated genes mapped strongly to the recycling endosome, exocytic vesicle membrane, synaptic vesicle membrane, and cortical endoplasmic reticulum, that could suggest increased vesicle trafficking and membrane turnover. The presence of motile cilium and peroxisome further highlights involvement in signalling and metabolic microdomains. The MF analysis revealed enrichment in transporter activities (amino acid transporters, metal cation antiporters) and regulatory elements such as kinase activator activity, PDZ domain binding, and actin binding. The simultaneous activation of proliferative and transport-related pathways, alongside cytoskeletal reorganization, could be suggestive of upregulated circRNA networks driving growth-oriented and invasive cellular programs in LUAD.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe GO analysis of downregulated circRNA-associated genes showed a contrasting enrichment pattern dominated by processes such as dual regulation of DNA-templated transcription, gene expression regulation, RNA polymerase II regulation, regulation of apoptotic processes, and intracellular signalling moderation, which could be indicative of attenuation of multiple regulatory checkpoints governing transcription and cell death (\u003c/span\u003e \u003cspan type=\"BoldSmallCaps\" class=\"BoldSmallCaps\" name=\"Emphasis\"\u003eSupplementary Fig.\u0026nbsp;2\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003e). Suppression of pathways controlling cell population proliferation and intracellular signal transduction suggests reduced regulatory constraint on growth and signalling fidelity. The key enriched CC terms included focal adhesion, cell substrate junction, ER membrane, mitochondrial membrane, and nuclear lumen, highlighting downregulation within adhesion structures, energy-regulating organelles, and nuclear regulatory spaces. In the MF domain, key downregulated terms involved DNA-binding transcription activator activity, cis-regulatory region binding, kinase binding, serine/threonine kinase activity, and GTPase regulator activity. All of which could be suggestive of decreased negative regulation of kinase pathways and transcriptional control mechanisms.\u003c/span\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e\u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eEnrichment of significant pathways\u003c/span\u003e\u003c/h2\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eA comparative KEGG pathway analysis of target genes regulated by dysregulated circRNA-miRNA interaction revealed a distinct division of labour between upregulated and downregulated circRNAs-miRNA networks, which coordinate unique but complementary activities that might contribute to cancer progression.\u003c/span\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eUpregulated circRNAs showed predominant associations with several disease- and metabolism-related pathways. Enriched pathways included Herpes simplex virus 1 infection, pathogenic E. coli infection, and Shigellosis, indicating activation of infection-responsive signalling modules. Additional enrichment was observed in pathways related to fluid shear stress and atherosclerosis, cellular senescence, gastric cancer, breast cancer, and the TGF-β signalling pathway, suggesting involvement of stress responses and cancer-linked signalling cascades. Metabolic pathways such as glycolysis/gluconeogenesis, RNA degradation, and amino-acid metabolism (glycine-serine-threonine metabolism and phenylalanine metabolism) were also significantly overrepresented. Pathways related to cytoskeletal regulation, including regulation of actin cytoskeleton, appeared among the enriched terms as well. Together, these enriched pathways indicate that upregulated circRNA-associated genes may support LUAD progression by enhancing metabolic activity, stress adaptation, and oncogenic signalling programs.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eIn contrast, downregulated circRNAs showed active associations with classical oncogenic signalling cascades representing of multiple signalling, cancer-associated, and hormonal regulation pathways. Enriched pathways included the cAMP signalling pathway, cGMP-PKG signalling pathway, and the MAPK signalling pathway, indicating reduced activity across major intracellular signalling routes. Several cancer-linked pathways were present, including pathways in cancer, proteoglycans in cancer, colorectal cancer, and renal cell carcinoma. Additional downregulated pathways involved ErbB signalling, adherens junction, and axon guidance, reflecting reduced signalling coordination and structural regulation. Hormonal and endocrine-associated pathways such as aldosterone synthesis and secretion, growth hormone synthesis, secretion and action, and the longevity regulating pathway also appeared among the enriched terms. Viral response pathways, including human cytomegalovirus infection, were similarly represented.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe collectively downregulated pathways signify diminished signalling fidelity, compromised structural and adhesion mechanisms, and the suppression of regulatory cancer-associated modules, which may jointly facilitate LUAD progression by undermining cellular control systems that typically inhibit malignant behaviour.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eMounting evidence have suggested that circRNAs, with their remarkable stability and regulatory versatility, predictive and diagnostic biomarkers for various forms of cancers\u003c/span\u003e [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Yet, the exact intricate interplay between dysregulated circRNAs, their target miRNAs, and downstream gene networks \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eremains largely ambiguous, leaving a critical gap in our comprehension of how these molecules shape tumour behaviour. Addressing this gap requires not only molecular characterization but also rigorous biomarker validation. This study offers an extensive and multidimensional exploration of the circRNA regulatory landscape in lung adenocarcinoma, revealing how deeply these non-coding transcripts are interwoven into the molecular architecture of tumour development and progression. CircRNAs have long remained an underexplored component of cancer biology, yet their exceptional stability, tissue-specific expression, and intricate interactions with miRNAs and RBPs position them as crucial regulators rather than mere transcriptional by-products. Our findings reinforce this emerging paradigm by demonstrating that circRNA dysregulation in LUAD is strongly associated with widespread alterations in transcriptional activity, signalling dynamics, cellular remodelling, and metabolic adaptation.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eUsing a systematic workflow that combined differential expression profiling, circRNA-miRNA-mRNA network reconstruction, RBP interaction prediction, and functional enrichment analysis, we identified a diverse set of circRNAs exhibiting significant dysregulation between cancer vs normal tissues. To strengthen the reliability of these molecular insights, we employed a three-layered biomarker validation strategy that integrated statistical testing, unsupervised clustering, and supervised machine-learning based prediction. This triangulated approach enabled the identification of 34 robust circRNA candidates consistently validated across all methodological frameworks (\u003c/span\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cspan type=\"BoldItalicSmallCaps\" class=\"BoldItalicSmallCaps\" name=\"Emphasis\"\u003e)\u003c/span\u003e. \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eUnlike single-method analyses that often produce context-dependent or noisy biomarker lists, this consensus strategy substantially enhances diagnostic confidence and translational potential.\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe global circRNA profiling of LUAD revealed a broad and structurally diverse circular transcriptome, comprising both known and novel circRNAs, with exon-derived species forming the predominant class. This distribution aligns with established circRNA biogenesis patterns in cancer, while the presence of intronic, overlapping, non-coding, and antisense circRNAs highlights additional regulatory layers that may be disrupted during malignant transformation. Differential expression analysis demonstrated substantial dysregulation, marked by a higher proportion of downregulated circRNAs compared to upregulated ones. This trend suggests widespread loss of epithelial identity and transcriptional repression in tumour tissue, whereas the upregulated subset likely represents circRNAs that support proliferative and oncogenic programs. The stark contrast between tumour and normal samples in PCA and heatmap analyses further highlight the distinct circRNA expression landscape characteristic of LUAD. Several of the top dysregulated circRNAs originated from genes involved in focal adhesion dynamics, extracellular matrix remodelling, cytoskeletal regulation, metabolic pathways, and chromatin or phosphatase-mediated control, biological processes well established in cancer progression. Notably, multiple circRNAs reported here, including those derived from PTK2, MMP1, KRT17, and CYP24A1 have been previously implicated in other malignancies, supporting their broader relevance as tumour-associated circular transcripts\u003c/span\u003e [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e], [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e], [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. This recurrence across cancer types strengthens the likelihood that these circRNAs serve conserved roles in growth signalling, invasion, and stress adaptation, and other fundamental cancer \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003ehallmarks\u003c/span\u003e [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e], [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e] .\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eAnalysis of the miRNA interaction networks revealed distinct regulatory architectures associated with upregulated and downregulated circRNAs in LUAD. The upregulated circRNAs formed a relatively compact but strategically focused network, with circRNAs such as hsa_circ_0035015, hsa_circ_0043614, and hsa_circ_0043632 functioning as prominent hubs by engaging multiple miRNAs. This concentrated interaction pattern suggests that a limited subset of highly expressed circRNAs may exert disproportionate regulatory influence. Conversely, the downregulated circRNAs constituted a more expansive and intricate network. Key circRNAs, including hsa_circ_0049036, hsa_circ_0046727, hsa_circ_0012248, and hsa_circ_0092361, interacted with a broad repertoire of miRNAs, indicating wider post-transcriptional disruption. While several miRNAs, e.g., hsa-miR-197-5p, hsa-miR-1285-3p, and hsa-miR-619-5p were observed to be recurrently targeted by multiple circRNAs, underscoring their centrality within the regulatory landscape. Collectively, these patterns reflect differential modes of network modulation associated with circRNA dysregulation in LUAD.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe circRNA-RBP interaction network in LUAD reveals a highly structured regulatory architecture involving 27 dysregulated circRNAs and 24 RNA-binding proteins with distinct binding patterns. Upregulated circRNAs exhibited substantially higher RBP connectivity, with hsa_circ_0005352 emerging as the principal hub interacting with 17 RBPs, followed by hsa_circ_0085740 and hsa_circ_0044130. Among RBPs, EIF4A3 displayed the most extensive binding spectrum, engaging 22 circRNAs across both expression groups, while AGO2 and HuR also interacted broadly, underscoring their roles as central regulators within the post-transcriptional network. The IGF2BP family demonstrated selective affinity toward upregulated circRNAs, particularly hsa_circ_0005352, hsa_circ_0044130, and hsa_circ_0085740, consistent with their established functions in stabilizing oncogenic RNAs. In contrast, METTL3 and DGCR8 interacted exclusively with downregulated circRNAs, suggesting that impaired processing or maturation of certain circRNAs may contribute to their reduced abundance in tumour samples. Additional RBPs, including C22ORF28, EWSR1, TAF15, and select LIN28 and FXR family members, preferentially bound upregulated circRNAs, whereas U2AF65 and ZC3H7B were enriched among interactions with downregulated species.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThis comprehensive circRNA-RBP interaction network reveals that dysregulated circRNAs in lung cancer exhibit non-random, expression-dependent RBP binding patterns, with upregulated circRNAs demonstrating enhanced multi-RBP associations that likely confer increased molecular stability and oncogenic functionality. The identification of EIF4A3, AGO2, and HuR as master regulatory hubs targeting both circRNA cohorts, alongside the selective engagement of IGF2BP family members with upregulated circRNAs and METTL3/DGCR8 with downregulated species, suggests that aberrant circRNA-RBP interactions constitute a fundamental mechanism driving lung cancer progression by rewiring post-transcriptional regulatory networks that control tumour cell proliferation, survival, metastasis, and therapeutic resistance.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe final panel of 34 high-confidence circRNAs, supported by GO and KEGG enrichment results, converged on key pathways linked to transcriptional regulation, EMT, metabolic reprogramming, and invasive signalling. Collectively, these findings indicate that LUAD progression is accompanied by coordinated disruption of circRNA networks, characterized by loss of regulatory circRNAs and selective amplification of those promoting proliferation, metabolic flexibility, and tumour aggressiveness.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe functional characterization of circRNA-miRNA-mRNA regulatory axes through GO and KEGG enrichment analyses offers critical insights into the molecular mechanisms underlying LUAD progression. By examining upregulated and downregulated gene sets independently, the analysis uncovers direction-specific biological alterations driven by circRNA dysregulation, thereby clarifying their distinct contributions to tumour behaviour. This integrative framework delineates the biological processes, cellular structures, and molecular functions most influenced by circRNA-mediated regulation, while parallel KEGG pathway enrichment broadens the understanding of how these genes converge on cancer-relevant signalling networks.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eUpregulated circRNA-associated genes demonstrated enrichment for functions related to elevated transcriptional activity, epithelial remodelling, vesicle trafficking, and cytoskeletal reorganization. Prominent GO terms such as DNA-templated transcription, epithelial proliferation, cell-cell adhesion, and neuron projection morphogenesis highlight enhanced transcriptional output, structural plasticity, and increased membrane dynamics. Correspondingly, enriched cellular components such as recycling endosomes, exocytic vesicle membranes, synaptic vesicles, and cortical ER suggest intensified intracellular transport and signalling turnover. Molecular function enrichment in transporter activity, kinase activation, PDZ-domain binding, and actin-binding reflects heightened metabolic and regulatory demands associated with tumour expansion and invasion. Together, these observations point toward upregulated circRNA networks promoting proliferative, invasive, and metabolically adaptive programs in LUAD.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eIn contrast, downregulated circRNA-associated genes were linked to repression of transcriptional regulators, apoptotic mediators, adhesion structures, and signalling modulators. Key biological processes including attenuated gene expression control, reduced apoptotic regulation, and suppressed intracellular signalling indicating weakening of homeostatic and anti-tumour safeguards. The downregulation of focal adhesions, mitochondrial and ER membranes, and nuclear regulatory domains further highlights compromised structural integrity and organelle-associated regulation. Reduced DNA-binding transcription activator activity, kinase binding, and GTPase regulatory functions also suggest a loss of regulatory precision in signal transduction.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eThe KEGG pathway analysis shows a strong functional difference between upregulated and downregulated circRNA-miRNA regulatory networks in LUAD. The pathways linked to upregulated circRNAs regulated metabolic activity, cellular stress adaptation, immune-responsive signalling, and cytoskeletal remodelling. These also included infection-related pathways, senescence, cancer-associated cascades, TGF-β signalling, glycolysis/gluconeogenesis, amino-acid metabolism, and regulation of the actin cytoskeleton, collectively reflecting a shift toward pathways that support tumour growth, plasticity, and survival. Conversely, downregulated circRNAs were associated with the suppression of key oncogenic and regulatory signalling systems. Attenuated pathways included cAMP, cGMP-PKG, MAPK, ErbB signalling, adherens junctions, axon guidance, and multiple cancer-linked pathways, along with endocrine regulatory modules such as aldosterone and growth hormone synthesis. The involvement of viral response and structural pathways further indicates a reduction in cellular coordination, adhesion, and signalling fidelity.\u003c/span\u003e \u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eOverall, these complementary patterns suggest that LUAD progression is facilitated by the simultaneous activation of pro-tumorigenic metabolic and signalling programs and the weakening of critical regulatory, structural, and homeostatic pathways highlighting the central role of dysregulated circRNA\u0026ndash;miRNA networks in reshaping the LUAD molecular landscape.\u003c/span\u003e \u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eIn summary, this work identifies 34 consistently dysregulated circRNAs in LUAD through differential expression patterns, regulatory network mapping, and multi-layer validation. Our three-pronged analytical approach enhances biomarker identification in circRNA research through deep learning, clustering, and statistical methods. The consensus across all methods for candidates such as hsa_circ_0058622 (SP100) and hsa_circ_0085740 (PTK2) provides strong support that these signals are biologically meaningful rather than artifacts of sequencing noise. Furthermore, several high-ranking circRNAs such as hsa_circ_0043632 (circKRT17), hsa_circ_0085751 and hsa_circ_0085740 (circPTK2), hsa_circ_0024109 (circMMP1), and hsa_circ_0060931 (CYP24A1) have been previously reported in other cancer contexts, offering independent corroboration for their functional relevance. Functional enrichment analyses highlight their involvement in transcriptional regulation, epithelial remodelling, metabolic adaptation, and cancer-associated signalling pathways, while predicted miRNA and RBP interactions further illuminate their possible regulatory roles. These integrated insights nominate several circRNAs as strong biomarker and therapeutic candidates. Collectively, these circRNAs hold promise for early detection, prognosis, treatment stratification, prediction of therapeutic response, and even development of RNA-based interventions such as antisense oligonucleotides or circRNA mimics. However, since this study is entirely based on computational prediction model, further experimental validation is essential to confirm their expression patterns, molecular interactions, and functional relevance before advancing toward clinical or translational applications.\u003c/span\u003e \u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgement:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors acknowledge the Bioinformatics Centre at Department of Bioscience and Biotechnology at Banasthali Vidyapith supported by DBT, Government of India, and the DST-FIST program of Government of India.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors received no financial support for the research, authorship, and/or publication of this article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors’ contributions:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAM was responsible for data collection and curation, conducting the analyses, visualizing the results, and preparing and extensively revising the initial manuscript draft. RB contributed to the conceptualization of the study, supervised the research, provided computational resources, and participated in the critical review and editing of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interests:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have declared that no conflict of interest exists.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll datasets analyzed in this study are publicly available in the repositories and databases cited within the Methodology section. While the technical context is provided via accession numbers in the text, direct URLs to the datasets are provided below to facilitate immediate access\u003c/p\u003e\n\u003cp\u003ehttps://www.ebi.ac.uk/ena/browser/view/SRP335547\u003c/p\u003e\n\u003cp\u003ehttps://www.ebi.ac.uk/ena/browser/view/SRP089923\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ehttps://www.ebi.ac.uk/ena/browser/view/SRP048484\u003c/p\u003e\n\u003cp\u003ehttps://www.ebi.ac.uk/ena/browser/view/SRP047399\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ehttps://www.circbase.org/\u003c/p\u003e\n\u003cp\u003ehttps://www.mirbase.org/\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis study did not involve human participants or animals; therefore, ethical approval is not required.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eChaitanya Thandra K, Barsouk A, Saginala K, Aluru JS, Barsouk A. Epidemiology of lung cancer. Wsp\u0026oacute;łczesna Onkologia. 2021;25(1):45\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5114/wo.2021.103829\u003c/span\u003e\u003cspan address=\"10.5114/wo.2021.103829\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSausville LN, Jones CC, Aldrich MC, Blot WJ, Pozzi A, Williams SM. Genetic variation in the eicosanoid pathway is associated with non-small-cell lung cancer (NSCLC) survival. PLoS ONE. Jul. 2017;12(7):e0180471. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0180471\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0180471\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSeguin L, Durandy M, Feral CC. Lung Adenocarcinoma Tumor Origin: A Guide for Personalized Medicine, \u003cem\u003eCancers (Basel)\u003c/em\u003e, vol. 14, no. 7, p. 1759, Mar. 2022, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cancers14071759\u003c/span\u003e\u003cspan address=\"10.3390/cancers14071759\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Z, Fillmore CM, Hammerman PS, Kim CF, Wong K-K. Non-small-cell lung cancers: a heterogeneous set of diseases. Nat Rev Cancer. Aug. 2014;14(8):535\u0026ndash;46. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrc3775\u003c/span\u003e\u003cspan address=\"10.1038/nrc3775\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen L-L. The biogenesis and emerging roles of circular RNAs. Nat Rev Mol Cell Biol. Apr. 2016;17(4):205\u0026ndash;11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrm.2015.32\u003c/span\u003e\u003cspan address=\"10.1038/nrm.2015.32\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCocquerelle C, Mascrez B, H\u0026eacute;tuin D, Bailleul B. Mis-splicing yields circular RNA molecules. FASEB J. Jan. 1993;7(1):155\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1096/fasebj.7.1.7678559\u003c/span\u003e\u003cspan address=\"10.1096/fasebj.7.1.7678559\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJeck WR, et al. Circular RNAs are abundant, conserved, and associated with ALU repeats. RNA. Feb. 2013;19(2):141\u0026ndash;57. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1261/rna.035667.112\u003c/span\u003e\u003cspan address=\"10.1261/rna.035667.112\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSalzman J, Chen RE, Olsen MN, Wang PL, Brown PO. Cell-Type Specific Features of Circular RNA Expression, \u003cem\u003ePLoS Genet\u003c/em\u003e, vol. 9, no. 9, p. e1003777, Sep. 2013, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pgen.1003777\u003c/span\u003e\u003cspan address=\"10.1371/journal.pgen.1003777\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalviya A, Bhuyan R. The recent advancements in circRNA research: From biogenesis to therapeutic interventions. Pathol Res Pract. Aug. 2023;248:154697. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.prp.2023.154697\u003c/span\u003e\u003cspan address=\"10.1016/j.prp.2023.154697\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeinonen R, Sugawara H, Shumway M. The Sequence Read Archive. Nucleic Acids Res. Jan. 2011;39:D19\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gkq1019\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkq1019\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Database.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarrison PW, et al. The European Nucleotide Archive in 2020. Nucleic Acids Res. Jan. 2021;49:D82\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gkaa1028\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkaa1028\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKent WJ et al. Jun., The Human Genome Browser at UCSC, \u003cem\u003eGenome Res\u003c/em\u003e, vol. 12, no. 6, pp. 996\u0026ndash;1006, 2002, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/gr.229102\u003c/span\u003e\u003cspan address=\"10.1101/gr.229102\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGlažar P, Papavasileiou P, Rajewsky N. CircBase: A database for circular RNAs, \u003cem\u003eRNA\u003c/em\u003e, vol. 20, no. 11, pp. 1666\u0026ndash;1670, Nov. 2014, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1261/rna.043687.113\u003c/span\u003e\u003cspan address=\"10.1261/rna.043687.113\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKozomara A, Birgaoanu M, Griffiths-Jones S. miRBase: from microRNA sequences to function. Nucleic Acids Res. Jan. 2019;47:D155\u0026ndash;62. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gky1141\u003c/span\u003e\u003cspan address=\"10.1093/nar/gky1141\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. no. D1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. Aug. 2014;30(15):2114\u0026ndash;20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/bioinformatics/btu170\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btu170\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGao Y, Wang J, Zhao F. CIRI: an efficient and unbiased algorithm for de novo circular RNA identification. Genome Biol. Jan. 2015;16(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13059-014-0571\u0026ndash;3\u003c/span\u003e\u003cspan address=\"10.1186/s13059-014-0571\u0026ndash;3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H, Durbin R. Fast and accurate short read alignment with Burrows\u0026ndash;Wheeler transform. Bioinformatics. Jul. 2009;25(14):1754\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/bioinformatics/btp324\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btp324\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePertea M, Pertea GM, Antonescu CM, Chang T-C, Mendell JT, Salzberg SL. StringTie enables improved reconstruction of a transcriptome from RNA-seq reads, \u003cem\u003eNat Biotechnol\u003c/em\u003e, vol. 33, no. 3, pp. 290\u0026ndash;295, Mar. 2015, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nbt.3122\u003c/span\u003e\u003cspan address=\"10.1038/nbt.3122\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLove MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. Dec. 2014;15:550. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13059-014-0550\u0026ndash;8\u003c/span\u003e\u003cspan address=\"10.1186/s13059-014-0550\u0026ndash;8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiranda KC et al. Sep., A Pattern-Based Method for the Identification of MicroRNA Binding Sites and Their Corresponding Heteroduplexes, \u003cem\u003eCell\u003c/em\u003e, vol. 126, no. 6, pp. 1203\u0026ndash;1217, 2006, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.cell.2006.07.031\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2006.07.031\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShannon P et al. Nov., Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks, \u003cem\u003eGenome Res\u003c/em\u003e, vol. 13, no. 11, pp. 2498\u0026ndash;2504, 2003, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/gr.1239303\u003c/span\u003e\u003cspan address=\"10.1101/gr.1239303\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDudekula DB, Panda AC, Grammatikakis I, De S, Abdelmohsen K, Gorospe M. CircInteractome: A web tool for exploring circular RNAs and their interacting proteins and microRNAs. RNA Biol. Jan. 2016;13(1):34\u0026ndash;42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/15476286.2015.1128065\u003c/span\u003e\u003cspan address=\"10.1080/15476286.2015.1128065\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRu Y, et al. The multiMiR R package and database: integration of microRNA\u0026ndash;target interactions along with their disease and drug associations. Nucleic Acids Res. Sep. 2014;42:e133\u0026ndash;133. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gku631\u003c/span\u003e\u003cspan address=\"10.1093/nar/gku631\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRiffo-Campos \u0026Aacute;, Riquelme I, Brebi-Mieville P. Tools for Sequence-Based miRNA Target Prediction: What to Choose? \u003cem\u003eInt J Mol Sci\u003c/em\u003e, vol. 17, no. 12, p. 1987, Dec. 2016. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijms17121987\u003c/span\u003e\u003cspan address=\"10.3390/ijms17121987\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen EY, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. Dec. 2013;14(1):128. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1471-2105-14\u0026ndash;128\u003c/span\u003e\u003cspan address=\"10.1186/1471-2105-14\u0026ndash;128\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. Jan. 2017;45:D353\u0026ndash;61. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gkw1092\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkw1092\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalviya A, Bhuyan R. Circular RNAs in cancer: roles, mechanisms, and therapeutic potential across colorectal, gastric, liver, and lung carcinomas. Discover Oncol. Jan. 2025;16(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s12672-025-01743\u0026ndash;9\u003c/span\u003e\u003cspan address=\"10.1007/s12672-025-01743\u0026ndash;9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang L, et al. Circular RNA hsa_circ_0008305 (circPTK2) inhibits TGF-β-induced epithelial-mesenchymal transition and metastasis by controlling TIF1γ in non-small cell lung cancer. Mol Cancer. Sep. 2018;17(1):140. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12943-018-0889\u0026ndash;7\u003c/span\u003e\u003cspan address=\"10.1186/s12943-018-0889\u0026ndash;7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang H, et al. CircPTK2 (hsa_circ_0005273) as a novel therapeutic target for metastatic colorectal cancer. Mol Cancer. Jan. 2020;19(1):13. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12943-020-1139\u0026ndash;3\u003c/span\u003e\u003cspan address=\"10.1186/s12943-020-1139\u0026ndash;3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDai W et al. Aug., CircMMP1 promotes colorectal cancer growth and metastasis by sponging miR\u0026ndash;1238 and upregulating MMP family expression, \u003cem\u003eAnn Transl Med\u003c/em\u003e, vol. 9, no. 16, pp. 1341\u0026ndash;1341, 2021, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/atm\u0026ndash;21\u003c/span\u003e\u003cspan address=\"10.21037/atm\u0026ndash;21\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u0026ndash;3930.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJi Y, Zhao Q, Feng W, Peng Y, Hu B, Chen Q. N6-Methyladenosine Modification of CIRCKRT17 Initiated by METTL3 Promotes Osimertinib Resistance of Lung Adenocarcinoma by EIF4A3 to Enhance YAP1 Stability, \u003cem\u003eCancers (Basel)\u003c/em\u003e, vol. 14, no. 22, p. 5582, Nov. 2022, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cancers14225582\u003c/span\u003e\u003cspan address=\"10.3390/cancers14225582\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu X et al. Feb., CircCYP24A1 hampered malignant phenotype of renal cancer carcinoma through modulating CMTM\u0026ndash;4 expression via sponging miR\u0026ndash;421, \u003cem\u003eCell Death Dis\u003c/em\u003e, vol. 13, no. 2, p. 190, 2022, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41419-022-04623\u0026ndash;0\u003c/span\u003e\u003cspan address=\"10.1038/s41419-022-04623\u0026ndash;0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi R, Chai L, Lei L, Guo R, Wen X. \u0026thinsp;MiR\u0026ndash;671\u0026ndash;5p sponging activity of \u0026lt;\u0026thinsp;scp\u0026gt;circMMP1 promotes esophageal squamous cancer progression, \u003cem\u003eThorac Cancer\u003c/em\u003e, vol. 14, no. 29, pp. 2924\u0026ndash;2933, Oct. 2023, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/1759\u0026ndash;7714.15078\u003c/span\u003e\u003cspan address=\"10.1111/1759\u0026ndash;7714.15078\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZou Q, Zhang Y, Zhu D, Liu X, Wang C, Xiang H. CircMMP11 as a prognostic biomarker mediates miR\u0026ndash;361\u0026ndash;3p/HMGB1 axis to accelerate malignant progression of hepatocellular carcinoma. Open Med. Nov. 2023;18(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1515/med\u0026ndash;2023\u0026ndash;0803\u003c/span\u003e\u003cspan address=\"10.1515/med\u0026ndash;2023\u0026ndash;0803\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. Jan. 2022;12(1):31\u0026ndash;46. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1158/2159\u0026ndash;8290.CD\u0026ndash;21\u003c/span\u003e\u003cspan address=\"10.1158/2159\u0026ndash;8290.CD\u0026ndash;21\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoizen MF. Hallmarks of Cancer: The Next Generation, \u003cem\u003eYearbook of Anesthesiology and Pain Management\u003c/em\u003e, vol. 2012, p. 13, Jan. 2012. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.yane.2012.02.046\u003c/span\u003e\u003cspan address=\"10.1016/j.yane.2012.02.046\" 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":false,"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":"circRNAs, LUAD, Machine Learning, Biomarkers, Precision Oncology","lastPublishedDoi":"10.21203/rs.3.rs-8684357/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8684357/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eLung cancer remains the leading cause of cancer-related deaths worldwide, with lung adenocarcinoma (LUAD) as its most prevalent subtype. Circular RNAs (circRNAs), known for acting as microRNA sponges and interacting with RNA-binding proteins, have emerged as key regulators in cancer biology. In this study, we introduce an integrated framework combining transcriptome profiling, network analysis, and machine learning to systematically identify and prioritise potential circRNA biomarkers in LUAD. We analysed RNA-seq datasets from LUAD samples and identified 52,744 circRNAs (18,922 novel and 33,822 previously known). Differential expression analysis revealed 1,480 significantly dysregulated circRNAs (855 downregulated and 625 upregulated) between tumour and normal tissues. To overcome limitations of traditional single-parameter screening, we implemented a three-pronged machine learning strategy integrating feature-weighted statistical scoring, unsupervised clustering with outlier detection, and deep neural networks. This multi-algorithm consensus approach identified 34 high-confidence circRNAs (17 upregulated and 17 downregulated) consistently prioritised across all methods, greatly reducing false positives. Functional enrichment analyses revealed distinct roles: upregulated circRNAs predominantly orchestrate metabolic reprogramming, epithelial\u0026ndash;mesenchymal transition, and cytoskeletal remodelling, whereas downregulated circRNAs govern transcriptional control, apoptosis regulation, and tumour-suppressor pathways. Key biomarker candidates include has_circ_0024109, hsa_circ_0058736, hsa_circ_0052195, hsa_circ_0085740, and hsa_circ_0060931 (upregulated), and hsa_circ_0016392, hsa_circ_0058622, hsa_circ_0012248, hsa_circ_0037308, and hsa_circ_0048053 (downregulated). This study provides a comprehensive, machine-learning-validated circRNA biomarker panel for LUAD, offering mechanistic insights into circRNA-driven oncogenesis and laying a foundation for next-generation circRNA-based diagnostics and therapeutics in precision oncology.\u003c/span\u003e \u003c/p\u003e","manuscriptTitle":"Integrated transcriptomic and machine learning-driven analysis reveals high-confidence circular RNA biomarkers in Lung Adenocarcinoma","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-19 11:24:05","doi":"10.21203/rs.3.rs-8684357/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":"eda1d501-6131-4a52-bd8d-1e795779eb6d","owner":[],"postedDate":"February 19th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-22T18:54:17+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-19 11:24:05","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8684357","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8684357","identity":"rs-8684357","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}