{"paper_id":"3c81969a-520a-40ba-9ddc-16b8bb35f033","body_text":"NAP: An Open-Source Pipeline for Cross-Domain Microbiome Profiling Using Nanopore Sequencing-Derived Amplicon Data | 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 software NAP: An Open-Source Pipeline for Cross-Domain Microbiome Profiling Using Nanopore Sequencing-Derived Amplicon Data Luke B. Jones, Stefan Bagby This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8173315/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background Nanopore sequencing offers a cost-effective and portable platform for amplicon-based microbiome analysis, but is still hindered by limited toolsets and sequencing error profile. While short-read technologies dominate microbial profiling workflows, their portability and flexibility are constrained. There is a need for robust pipelines tailored to Nanopore data that can support cross-kingdom ribosomal RNA profiling. Results We introduce the Nanopore sequencing-based Amplicon Pipeline (NAP; https://github.com/Luke-B-Jones/NAP ), an open-source tool optimised for flexible, mixed-domain primer sets (such as 515Y/926R). NAP performs quality filtering, chimera removal, centroid identification, and BLAST-based taxonomic classification with consensus correction. It outputs normalised, bias-corrected tab-separated value files suitable for downstream analysis. Validation against two commercial mock communities showed that NAP achieves genus-level precision of up to 100%, with taxonomic concordance comparable to Illumina-based workflows. Detection sensitivity was consistently reliable above 1% relative abundance. β-diversity measures, including Bray–Curtis and Jaccard indices, fell within expected replicate variation. Taxonomic agreement remained high across a range of read depths and sequencing qualities, with most errors attributable to laboratory-derived artefacts rather than computational limitations. Conclusions NAP delivers robust genus-level performance on par with Illumina workflows, with the potential to achieve species-level resolution using longer amplicons. Its compatibility with portable and cost-effective sequencing makes it well suited for accurate long-read microbiome profiling in both laboratory and field environments. Amplicon sequencing Microbiome Nanopore sequencing Ribosomal RNA Taxonomic classification Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Background Microbiomes are increasingly studied from a cross-domain perspective, capturing interactions among bacteria, archaea, and eukaryotes across diverse environmental and host-associated contexts. While microbiome research has historically focused on bacteria, recent work highlights that inter-domain interactions are central to understanding community structure and function, and health-related outcomes. In clinical settings, bacterial–fungal interactions have physical, metabolic, and immune implications in disease (Arvanitis and Mylonakis, 2015 ). Inter-kingdom microbial networks are now recognized as potential drivers of complex pathologies and may offer novel diagnostic and therapeutic targets (Li et al., 2018 ; Wang et al., 2023 ). Beyond medicine, these interactions shape ecological dynamics, influence food microbiology, and affect agricultural productivity (Frey-Klett et al., 2011 ). As such, accurate and domain-inclusive profiling of microbial communities is essential for advancing microbiome science across multiple disciplines. While shotgun and whole genome sequencing approaches offer high taxonomic resolution and low taxonomic bias, they remain costly and computationally intensive. For studies focused primarily on community composition, amplicon sequencing provides a faster and more cost-effective alternative. Traditionally, however, amplicon-based profiling has been restricted to bacteria using 16S rRNA primers, or has relied on separate 16S and 18S reactions to study bacterial and eukaryotic components independently. Because these are amplified and sequenced in isolation, their abundance outputs are not directly comparable, limiting cross-domain integration. Recent advances in primer design have addressed this issue by enabling the amplification of bacterial, archaeal, and eukaryotic small subunit rRNA from a single reaction using universal primer sets. Illumina-based workflows offer high base accuracy and are compatible with such primers, but remain limited by short read lengths. Moreover, while mature and user-friendly pipelines exist for Illumina data, these often require programming expertise and lack tools for domain-aware normalisation or bias correction. Nanopore sequencing, in contrast, supports real-time, full length amplicon sequencing with much lower capital outlay and greater portability (Lao et al., 2021 ; Charalampous et al., 2019 ). These features make it an appealing choice for field-based or resource-limited settings. However, existing Nanopore amplicon tools are largely limited to bacterial 16S pipelines (e.g., EPI2ME) or require researchers to build bespoke workflows from scratch, rendering them inaccessible to most potential users, particularly those aiming to study mixed-domain microbial communities. We present here the Nanopore sequencing-based Amplicon Pipeline (NAP), a lightweight, open-source pipeline purpose built to transform Nanopore amplicon sequencing data into high quality, bias-corrected, and taxonomically normalised microbiome profiles. Unlike existing pipelines which are either tailored exclusively for bacterial profiling or require extensive computational expertise, NAP is designed to accommodate any cross-domain rRNA primer sets (such as the default set, 515Y/926R) supporting simultaneous bacterial, archaeal, and eukaryotic profiling from a single amplicon pool. Implementation SILVA rRNA reference databases are pre-filtered to exclude ambiguous, unclassified, or contaminant entries (e.g., plant mitochondrial rRNA), and are split into 16S and 18S subsets. Primer sequences are then mapped to each entry, extracting the longest amplicon region per entry via ungapped local alignment; untrimmed entries are retained where necessary. These domain-specific trimmed databases are then indexed for BLAST-based classification. NAP processes demultiplexed amplicon reads (FASTQ) through a modular workflow optimised for mixed-domain microbiome profiling. Reads are first filtered dynamically, where read depth and quality are optimised by adjusting quality thresholds and masking low quality bases. Chimeric reads are then removed, and the remaining reads (termed the RAW dataset) are clustered into centroids (termed CNT dataset; Fig. 1 A and B). Centroid sequences are classified via BLAST against the custom SILVA reference database, and taxonomic identities are assigned using a majority consensus algorithm. These annotated CNT sequences then serve as a high-confidence internal database to reclassify all RAW reads against, boosting precision while reducing computational load (Fig. 1 C). Abundance data are normalised using total sum scaling (TSS), corrected for known 16S/18S domain bias, and filtered to remove contaminants and low confidence taxa. Final outputs include genus- and species-level abundance tables (.tsv), pipeline logs (.txt), and quality summaries to guide user confidence (Fig. 1 D). NAP is implemented in Bash and Python and is designed for Unix/Linux-based environments. It is operated entirely via the command line, with no programming required beyond Conda and installation. Users configure the pipeline via a main configuration file that sets key parameters such as quality thresholds, filtering behaviour, and reference paths. Primer-specific settings are handled via modular sub-configuration files, allowing users to easily adapt the workflow to different primer sets, experimental designs, and hardware usage without altering the core code. Installation instructions are available via GitHub ( https://github.com/Luke-B-Jones/NAP ). The pipeline demonstrates robust genus- and species-level classification accuracy, handles low depth and high noise data gracefully, and ensures reproducible outputs through consistent detection and strong agreement metrics. NAP thus fills a niche between highly accurate but capital- and infrastructure-intensive Illumina approaches and field-compatible emerging Nanopore workflows. By combining adaptability, accessibility, and analytical rigour, it unlocks the potential of Nanopore sequencing for rapid, accurate, and cross-domain microbiome analysis across contexts. Methods Mock Community Preparation Two commercially available mock microbiome communities were used to validate the pipeline: ZymoBIOMICS Microbial Community Standard II (Log Distribution), and ZymoBIOMICS Gut Microbiome Standard (both Cambridge Biosciences); henceforth, these are referred to as log mock (L1, L2, and L3) community and gut mock (M1, M2, and M3) community, respectively. In addition to this, environmental blanks were made by exposing sterile 0.9% saline to the laboratory environment. DNA was extracted from these samples using ZymoBIOMICS DNA/RNA miniprep kit according to the manufacturer’s recommendations (“ZymoBIOMICS DNA/RNA Miniprep Kit,” 2025). Homogenisation was done according to a custom protocol, with six cycles of 30 seconds at 9,000 rpm using a Bertin Precellys Evolution bead beater, increasing the yield of harder-to-lyse organisms (Zhang et al., 2020 ). DNA was quantified using a Qubit fluorometer (Thermo Fisher Scientific). Isolates were then amplified using Phusion™ Plus PCR Master Mixes (Thermo Fisher Scientific), using an annealing temperature of 50°C and ca. 100ng of template DNA per reaction, along with the recommended guidelines for 50 µL single reaction protocol outlined by the supplier, and custom oligonucleotide primers (Eurofins) (515Y, 5′- GTGYCAGCMGCCGCGGTAA , 926R, 5′- CCGYCAATTYMTTTRAGTTT; McNichol et al., 2021 ). All amplifications were validated using gel electrophoresis by identifying 16S and 18S bands at ca. 300–500 and ca. 700 bases, respectively. This resulted in six amplicon pools, three replicates per mock community. Amplicon pools were then sequenced using one MinION R10.4.1 flow cell according to the Native Barcoding Protocol (SQK-NBD114.24), and then basecalled and demultiplexed using Dorado (v0.9.0; “nanoporetech/dorado,” 2025; super accuracy model v5.0.0). Statistical Analysis Statistical analysis was performed using a custom Python script to evaluate the accuracy, reliability, and structural validity of the pipeline’s outputs relative to the known composition of two commercially available mock communities. Genus- and species-level taxonomic tables were generated from each sample. To enable comprehensive sensitivity profiling, taxa absent across all replicates were assigned zero counts. Detection accuracy was quantified by calculating precision, recall, and F1-scores for each replicate at a fixed detection threshold (τ = 0.001), using the expected genera as the reference set. To evaluate quantitative agreement between observed and expected taxon abundances, linear regression and Lin’s concordance correlation coefficient ( ρ CCC ) were used. This enabled both proportional and absolute agreement to be assessed. Agreement plots compared mean observed abundances across replicates against their expected values, while Bland–Altman plots quantified systematic bias and calculated limits of agreement using ± 1.96 standard deviations from the mean difference. Together, these analyses assessed not only detection but also abundance accuracy and consistency. This was restricted to taxa which were expected, meaning false positives were not included in this part of the analysis. Community-level similarity and sample structure were explored using β-diversity metrics. Pairwise dissimilarities were calculated using both Jaccard distance (based on presence/absence data) and Bray–Curtis dissimilarity (based on relative abundances). These dissimilarity matrices were visualised using principal coordinate analysis (PCoA) to identify clustering of replicates around their respective mock community centroids. A replicate was defined as a success if it was closer in β-diversity space to its own mock profile than to that of the alternate mock group. Statistical significance of replicate fidelity was assessed using one-sided binomial tests for each group, with groupwise p -values aggregated via Fisher’s method to test for global consistency. To investigate taxon-level variability, bar plots were generated showing the expected genus abundances overlaid with individual replicate values. For each genus, the coefficient of variation (CV) across replicates was annotated, with values exceeding 0.5 highlighted to indicate high within-group variability. Detection sensitivity was further examined by plotting the proportion of replicates that detected each expected taxon against its relative abundance, providing a measure of detection robustness at different abundance levels. These analyses included all observed taxa, including false positives. Results Read Filtering and Quality Control NAP applies an automated read level filtering step that dynamically adjusts Phred score thresholds to balance retained depth and base quality, thereby optimising downstream taxonomic accuracy. After filtering, the number of retained reads correlates positively with the initial input read count (Figure 2A). The observed variance around this trend is primarily attributable to sample-specific error profiles arising from variability in sequencing throughput, amplification efficiency, and the initial DNA integrity. To limit computational overhead, samples are deliberately downsampled to 225,000 reads, rather than allowing excessive retention after filtering; this is observed in the logarithmic mock repeats, with read counts being capped at 225,000 (Figure 2A). In most samples, the adaptive filtering maintained thresholds above Q30 (average Phred score of reads) while retaining over 100,000 high quality reads (Figures 2A, 2B). For lower quality samples such as M1, the pipeline adaptively reduced the threshold to Q24, masking bases below Q5 (corresponding to an error probability of ~31.6%, and representing less than 0.01% of all bases) rather than discarding more reads entirely. This preserved sufficient depth (>10,000 reads) while minimising erroneous taxonomic assignments. This was done to a lesser extent in higher depth samples; for example, mock L1 was masked below Q1 (corresponding to an error probability of ~79.4%, and affecting fewer than 0.001% of all bases). Chimera removal using VSEARCH led to the exclusion of 8.55% ± 3.37% of reads on average, meaning read depths were not substantially affected downstream. Overall, NAP’s default filtering effectively balanced read depth and quality, though user-defined parameter tuning may improve performance for datasets with atypical quality profiles. Pipeline Refinement Initial taxonomic assignments were refined to improve community structure fidelity and mitigate Nanopore-specific sequencing errors, including substitution and indel artefacts. Multiple refinement strategies were evaluated for recovering accurate amplicon sequence variants (ASVs). Methods included clustering-based approaches such as CD-HIT, isONclust, and Rattle to identify sequence centroids. In parallel, alignment-based consensus calling with Medaka, Racon, and transcript-style workflows were evaluated, using both reference-guided and centroid-guided strategies. Taxonomic classification methods were compared using Kraken2, QIIME2, and BLAST to assess accuracy and robustness across toolkits. Consensus sequence–polishing approaches (e.g., Medaka, Racon, and custom tools) were inconsistent, particularly for low abundance taxa, and frequently propagated sequencing errors into final profiles. In contrast, BLAST provided tuneable alignment parameters that could be optimised for Nanopore-specific error profiles. Our BLAST with CD-HIT configuration improved centroid-level taxonomic assignment accuracy relative to other methods. BLAST also enabled a post hoc consensus voting algorithm to efficiently resolve classification ambiguities at the species level. The speed of pipeline completion was improved by SILVA database refinement (trimming regions not amplified by primer set, and removing taxonomically unresolved entries e.g., uncultured, metagenomically assembled), and this increased consensus algorithm accuracy (partially due to refined regions, but also by removing ambiguous entries). Using this approach, classification rates ranged from 98.53% to 99.88% in >Q30 samples, and reached 94.68% in Q20 samples (species level); this is comparable to the performance of NG-Tax, an Illumina-based amplicon pipeline (Ramiro-Garcia et al ., 2018), under tested conditions. β-diversity analyses showed strong concordance between observed and expected profiles at both genus and species levels. For all mock communities, genus level Bray–Curtis and Jaccard distances were ≤ 0.5 (Figure 3). At the species level, the logarithmic mock community achieved Bray–Curtis distances ranging from 0.01 to 0.13 and Jaccard distances from 0.33 to 0.50. The gut mock community exhibited slightly higher but consistent Bray–Curtis distances (0.44–0.47) and Jaccard distances (0.33–0.45). In contrast, alternative refinement strategies tested during pipeline development frequently exceeded 0.6 for both metrics. These results support the final pipeline configuration, centred on CD-HIT clustering and BLAST-based classification with consensus correction, as the most accurate and reproducible method. The clear convergence toward expected taxonomic structure supports its effectiveness for Nanopore-based microbiome analysis (Figure 3). Taxonomic Classification All expected bacterial and eukaryotic genera were detected across both mock communities, with only Saccharomyces exhibiting high replicate variability (coefficient of variation > 0.5; Figure 4A). The only archaeal genus (Methanobrevibacter) was detected below the pipeline’s 0.05% confidence threshold. Similarly, Salmonella was detected but also fell below the applied filtering threshold. Only four genera were identified as false positives at the genus level: Shigella, likely reflecting misclassification due to the presence of multiple Escherichia strains and their known sequence similarity within the V4–V5 small subunit ribosomal RNA regions; and Bysmatrum, Pasteurellaceae, and Streptococcus, which displayed sporadic, highly variable abundances and were also present in blank controls, suggesting contamination beyond the detection limits of the current decontamination method. This interpretation is supported by previous reports identifying Lactobacillus, Saccharomyces, and Streptococcus as common laboratory- or human-associated contaminants (Glassing et al., 2016; Salter et al., 2014). Taxonomic accuracy was mostly preserved at the species level. However, Veillonella rogosae was not detected in any gut mock replicates; instead, artefactual misclassification occurred under V. dispar and V. parvus. A similar but less pronounced pattern was observed for Lactobacillus fermentum, which showed inflated abundance due to misattribution to L. acidophilus, a known and present contaminant. These cases are consistent with the high sequence similarity of the V4–V5 region among closely related taxa (Janda and Abbott, 2007). For both Veillonella and Lactobacillus, minor Nanopore sequencing errors likely contributed to taxonomic divergence during species level classification. At both genus and species levels, replicate profiles were statistically similar to the expected mock community composition (p < 0.05; Figure 4), indicating high reproducibility and classification accuracy. Most expected taxa were successfully identified with low inter-replicate variability (e.g., CV < 0.4 in most cases, often reaching < 0.15). Contaminants were identified based on prevalence in blank controls, high blank-to-sample abundance ratios, and exclusion from expected taxa lists. On average, the decontamination process removed 7.00 ± 2.68 species level hits per replicate, all of which were present in blank samples and judged to be contaminants following manual review. An additional 9.83 ± 6.49 species per replicate exhibited altered abundances after decontamination without being fully removed. For several high abundance taxa (such as Faecalibacterium prausnitzii, Listeria monocytogenes, and Pseudomonas aeruginosa), these adjustments moved observed abundances closer to expected values and reduced variability across replicates (Figure 4). Together, t hese findings indicate that, although the current decontamination strategy is relatively simple and does not fully resolve artefactual misclassification, it effectively identifies common contaminants and improves taxonomic accuracy. Classification performance is highly reliable at the genus level, and with improved primer design or longer read lengths, species level accuracy could be enhanced. Taxon Abundance Accuracy The pipeline’s default primers, 515Y/926R, target the V4–V5 region of the small subunit (SSU) rRNA and offer broad cross-domain coverage, amplifying ca. 96% of known rRNA sequences across bacteria, archaea, and eukaryotes. Although these primers improve upon earlier cross-domain designs, they still introduce bias through differential binding site specificity and annealing efficiency, which can skew taxon representation (Parada et al ., 2016; McNichol et al ., 2021). Such effects are further compounded by wet lab variables, particularly DNA extraction efficiency, and SSU copy number variation. To mitigate these biases, we implemented prolonged bead beating cycles to balance DNA recovery from both easily lysed Gram-negative bacteria and more resilient taxa, including Gram-positive bacteria, spore formers, and eukaryotes (Zhang et al ., 2020). We assessed the accuracy of relative abundance estimates by comparing observed taxon abundances against expected values from mock community profiles using agreement plots and Bland–Altman analysis at genus level (Figure 5). The logarithmic mock community showed excellent concordance, with minimal Bland–Altman bias (21.2 units) and strong statistical agreement across replicates ( r = 1, p = 1.5e -5 , p CCC = 1). In contrast, the gut mock community showed reduced correlation ( r = 0.68, p = 2.5e -3 , p CCC = 0.53) and a substantial negative bias (–5876.6), suggesting systematic underrepresentation of several expected taxa. However, intra-taxon variability remained low (Figure 4A), indicating that discrepancies were driven by consistent biases, potentially due to primer mismatch or extraction inefficiencies, rather than random noise or computational error. To further validate the pipeline’s reliability, we repeated the agreement and Bland–Altman analyses at species level (Figure 6). Results remained statistically robust, with only marginal reductions in correlation (e.g., gut mock: r = 0.68 genus vs. 0.66 species; p = 2.5×10⁻³ vs. 1.7×10⁻³; ρ CCC = 0.53 vs. 0.56) and reduced Bland–Altman bias in both mock communities (e.g., log mock: 21.2 genus vs. 17.7 species). These patterns suggest that, in well characterised taxa, species level profiles produced by the pipeline remain quantitatively trustworthy. Nonetheless, while species level abundances appear numerically more concordant, taxonomic resolution is often compromised by limitations of the V4–V5 region, which can fragment expected taxa into multiple spurious species (e.g., Veillonella being misassigned to two absent species). As such, genus level assignments remain more biologically reliable according to these results, and are therefore preferred for robust interpretation under the pipeline’s default settings. While 515Y/926R taxonomic bias has not been well explored, there is evidence that the observed abundance discrepancies are not attributable to pipeline error. The phylum Pseudomonadota (formerly Proteobacteria ) was well represented, with Pseudomonas showing expected abundances, as observed in other studies (Klindworth et al ., 2013). Conversely, as expected, Actinobacteriota was underrepresented, with Bifidobacterium displaying particularly low abundance. Amplicon bias is known in this phylum, and is potentially compounded here by the Gram-positive cell wall structure (Parada et al. , 2016). Bacillota (previously Firmicutes) were moderately underrepresented overall, while Bacteroidetes were more substantially affected, a pattern of particular interest given the clinical relevance of the ‘Firmicutes:Bacteroidetes’ ratio (Palkova et al ., 2021). Within these phyla, Faecalibacterium , Lactobacillus, Listeria , and Veillonella showed good or mildly reduced recovery, while Bacteroides and Prevotella were more strongly underrepresented. This skew was less pronounced than in some prior studies, possibly due to differential lysis efficiency: Gram-positive Bacillota may be more effectively extracted than Gram-negative Bacteroidetes under prolonged bead beating conditions (Parada et al. , 2016; Zhao et al ., 2023). Bacillus detection was inconsistent across replicates. In two gut mock community samples, it was present below the reporting threshold; in a third, it appeared overrepresented. As a spore forming, Gram-positive taxon, Bacillus is known to be sensitive to extraction and amplification biases. Additionally, Table 1 supports the possibility of mismatch-based primer bias, a key factor in under-amplification of rRNA. Furthermore, its presence in extraction blanks suggests potential cross-contamination, meaning its abundances are further reduced. Misalignment may also explain the inflated abundance in one replicate, given the high sequence similarity between Bacillus strains, where a single significant contamination event caused reinforcement of the taxon’s presence. A comparable pattern was seen with Shigella , which was not present in the mock community but appeared alongside Escherichia , likely due to shared rRNA sequence regions and minor contamination (Zhao et al ., 2023). Other taxa have poorly characterised biases for primer set 515Y/926R, limiting interpretation. Here, for example, Methanobrevibacter smithii and Salmonella enterica were both detected below the pipeline’s filtering threshold (0.05% relative abundance filtration threshold; 0.2% and 0.089% expected relative abundance for M. smithii and S. enterica, respectively), suggesting mild underrepresentation, with only Salmonella enterica showing potential for primer mismatches (Table 1). Similarly, Akkermansia muciniphila showed no primer mismatches and underrepresentation, with bias uncharacterised in the literature. For Methanobrevibacter smithii, the difficulty of archaeon lysis is a strong candidate contributor, but other factors can lead to related archaea being underrepresented (Youngblut et al., 2021; Zhao et al. , 2023). Akkermansia muciniphila and Salmonella enterica are easy to lyse Gram-negative, non-spore forming bacteria, suggesting excessive bead beating may have led to a slight underrepresentation bias, but this is likely compounded by other PCR-related biases, considering mismatches are not the only known factor in primer-induced bias (Qin et al. , 2023; Silverman et al ., 2021). Finally, as the 515Y/926R primers also capture eukaryotic 18S rRNA, a domain correction factor is required to adjust for lower 18S amplification efficiency. The pipeline applies a default correction factor of 0.4, based on empirical estimates ranging from 0.3–0.5. This correction was validated in our results: Candida albicans and Saccharomyces cerevisiae were detected at or near to the expected abundances relative to co-occurring 16S taxa (Yeh et al ., 2021). Taken together, replicate abundances were highly consistent, demonstrating strong technical reproducibility. Deviations from expected values were largely attributable to primer mismatch, variability in lysis efficiency, or low input abundance (rather than stochastic or computational error). The pipeline’s default 16S/18S correction proved effective, and the bead beating protocol did not seem to induce a systematic skew between Gram-positive and Gram-negative bacterial taxa. These findings support the pipeline’s capacity to generate biologically representative taxonomic profiles across complex microbiomes. Table 1: TestPrime Output Assessing Primer Bias Across Observed Genera. TestPrime was run using SSU SILVA version 138.2 RefNR (pruned to remove redundancy). Mismatches were configured to permit three mismatches, with no enforced 0-mismatch zone at the 3′ end. Results indicated 96.6% coverage of the SILVA database. The forward primer was responsible for mismatches in 50.6% of cases, and the reverse primer in 46.0% of cases, while mismatches occurred in both primers in 3.4% of cases; these results refer to an in silico evaluation of theoretical primer binding, and therefore do not account for other external factors. The \"Perfect Match\" column shows the percentage of references that aligned to an unambiguous primer version with zero mismatches. “Mean Pair Coverage” shows the literal primer-to-template (reference) alignment coverage across all best unambiguous primer-to-reference matches in a given taxon. “3′-end MM Forward” displays the percentage of reference–primer matches with at least one 3′-end mismatch for the forward primer, and “3′-end MM Reverse” shows the same for the reverse primer. Species Perfect Match (%) Mean Pair Coverage (%) 3’-end MM Forward (%) 3’-end MM Reverse (%) Akkermansia muciniphila 100.0 100.0 0.0 0.0 Bacillus subtilis 94.2 99.7 2.4 1.9 Bacteroides fragilis 100.0 100.0 0.0 0.0 Bifidobacterium adolescentis 100.0 100.0 0.0 0.0 Candida albicans 92.9 99.6 2.9 0.0 Clostridioides difficile 99.6 100.0 0.4 0.0 Escherichia coli 99.7 100.0 0.1 0.1 Faecalibacterium prausnitzii 100.0 100.0 0.0 0.0 Fusobacterium nucleatum 97.8 99.9 0.7 0.7 Lactobacillus fermentum 95.4 99.8 3.4 0.6 Listeria monocytogenes 99.3 100.0 0.4 0.1 Methanobrevibacter smithii 100.0 100.0 0.0 0.0 Prevotella corporis 100.0 100.0 0.0 0.0 Pseudomonas aeruginosa 95.7 99.8 1.7 1.5 Roseburia hominis 100.0 100.0 0.0 0.0 Saccharomyces cerevisiae 91.4 99.3 2.9 0.0 Salmonella enterica 99.7 100.0 0.1 0.1 Veillonella rogosae 100.0 100.0 0.0 0.0 Quantitative Performance Metrics Detection sensitivity and classification accuracy were assessed using the logarithmic and gut mock communities separately (Figure 7). In the logarithmic mock community, taxa above ca. 1% relative abundance were reliably detected across all replicates, while those below ca. 1% exhibited variable detection rates (Figure 7A). This indicates a steep detection drop off at low abundance levels, characteristic of primer and amplification bias in highly skewed communities. The gut mock community showed a similar trend, with taxa above ca. 1% relative abundance being reliably detected, but there were two instances of a single repeat failing to detect an organism (Figure 7B). Precision–recall analysis revealed strong classification performance across replicates. In the gut mock community, all replicates achieved high precision (>0.8) and recall (>0.8), with Q30+ samples (M2, M3) clustering near the identity line, indicating accurate and comprehensive taxon detection (Figure 7B). The lower quality replicate (M1, Q23) displayed a modest reduction in recall but maintained high precision, suggesting resilience of the classification strategy to reduced read quality. In contrast, the logarithmic mock community samples showed perfect precision (1.00) across all replicates, but substantially lower recall (0.50–0.70), reflecting the challenges of detecting very low abundance taxa rather than false positives (Figure 7A). Taken together, these results confirm that the pipeline delivers strong classification accuracy and consistent detection sensitivity across replicate samples. Genus level precision remained high even in lower quality data, with minimal false positives observed. Recall varied depending on taxon abundance and sample quality, with high abundance taxa (>1% relative abundance) reliably detected across all replicates, and a consistent drop off below this level, particularly in the more skewed logarithmic mock. The lowest confidently detected taxon in the gut mock community was present at 2.8% relative abundance and recovered across all replicates, while in the logarithmic mock community, a taxon detected at 0.089% was recovered in only one of three replicates. This suggests that the empirical limit of detection for this pipeline lies between 0.1% and 1% relative abundance, but this cannot be accurately defined from the available data; at ≥1% we observed high precision and recall across replicates. Together with precision and recall consistently >0.85 , performance is in line with established performance expectations for amplicon-based microbiome profiling (Poncheewin et al ., 2020). Moreover, the observed precision and recall values compare favourably with Illumina-based pipelines such as NG-Tax and QIIME2 (Poncheewin et al ., 2020), highlighting the pipeline’s reliability. These findings support the utility of this tool for accurate, reproducible, and domain-inclusive microbial community analysis across a range of sequencing conditions. Discussion Despite increasing interest in Nanopore sequencing-based microbiome profiling, streamlined and flexible pipelines for long read, cross-domain amplicon analysis are in short supply. Existing tools such as EPI2ME lack the taxonomic and primer adaptability needed for modern studies. NAP was developed to address these gaps, enabling high throughput, domain-inclusive rRNA amplicon profiling across bacteria, archaea, and eukaryotes. NAP emphasises genus level accuracy while supporting species level resolution through careful interpretation. Key features, such as automated 16S/18S correction, primer-specific database pruning, and user-friendly configuration, make NAP both adaptable to diverse primers and accessible to researchers with minimal coding expertise. Taxonomic dissimilarity metrics support the pipeline’s reliability. Genus level Bray–Curtis dissimilarities in the logarithmic mock community were consistently low (< 0.13), within the normal variation reported for technical replicates (Bray–Curtis: ~0.12 ± 0.04; Yeh et al., 2017 ). In the more complex gut mock community, Bray–Curtis values remained below 0.4, with higher variation attributable to sample-specific artefacts and known wet lab limitations. Jaccard dissimilarities, which are expected to be higher due to presence/absence sensitivity, typically remained < 0.33 across the mock communities, well within the 0.2–0.5 range reported as expected variance between other state-of-the-art pipelines (as a proxy for the variance expected within a single pipeline’s outputs; O’Sullivan et al., 2021 ). NAP demonstrated robust performance across sequencing conditions. Its adaptive filtering allowed accurate classification even in lower quality samples (e.g., Q20, < 100,000 reads), while high quality samples (Q30+, > 100,000 reads) produced highly consistent outputs. Reproducibility across technical replicates was high, with correlation coefficients (r ranging 0.68 to 1.0), concordance ( p CCC ranging 0.53 to 1.0), and precision (1.0 in logarithmic, ca. 0.85 in gut mocks) aligning with benchmarks from Illumina-based tools (Poncheewin et al., 2020 ) and, in some cases, even approaching short read metagenomics performance (Adams et al., 2023 ). Recall in logarithmic mock communities was lower ( ca. 0.5–0.7), but this is primarily attributable to primer bias and lysis inefficiencies, recognised limitations in amplicon workflows, rather than classifier error. Relative abundance discrepancies were primarily driven by wet lab constraints. Underrepresentation of certain taxa (e.g., Akkermansia and Bifidobacterium ) aligned with known primer mismatches and lysis challenges. Species level artefacts were rare and traceable to either contaminating taxa or closely related sequences (e.g., Escherichia–Shigella ). The pipeline’s conservative decontamination strategy eliminated most known contaminants, and corrected true positive abundances appropriately when inflated by contamination. In conclusion, NAP provides an accurate, reproducible, and open source solution for long read microbiome profiling. Its performance at the genus level is comparable to Illumina pipelines, under tested conditions, and species level accuracy is achievable with appropriate interpretation. Combined with its modular structure, primer-aware database integration, and low technical barrier, NAP fills a critical need in Nanopore amplicon sequencing workflows, supporting robust microbiome analysis across bacterial, archaeal, and eukaryotic domains, with modular components allowing primer-specific customisation. Abbreviations ASV amplicon sequence variant bp base pair BLAST basic local alignment search tool CD-HIT cluster database at high identity with tolerance CV coefficient of variation DNA deoxyribonucleic acid PCoA principal coordinate analysis PCR polymerase chain reaction Q-score (Phred score) quality score representing probability of error in base calling rRNA ribosomal ribonucleic acid SSU small subunit (of rRNA) TSS total sum scaling TSV tab-separated values ρc (ρCCC) Lin’s concordance correlation coefficient, β-diversity:beta diversity Declarations Availability of data and materials The datasets and code supporting the conclusions of this article are available as follows: Project name: NAP Project home page: https://github.com/Luke-B-Jones/NAP Archived version: A permanently archived release is available via Zenodo at DOI: 10.5281/zenodo.17662789 Operating system(s): Platform-independent (tested on Linux) Programming language: Python (≥3.9) and Bash Other requirements: Standard UNIX environment, and conda License: MIT license (open-source; free for academic and commercial use) Restrictions to use by non-academics: None. The raw Nanopore amplicon sequencing data generated during this study are available in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA1367045. Competing interests Luke Jones’s Ph.D. studentship is partly funded by Oxford Nanopore Technologies plc., which had no role in study design, data analysis, or manuscript preparation. Funding Work funded by the University of Bath and Oxford Nanopore Technologies plc. Authors' contributions LJ designed the study and implemented the pipeline, and carried out analyses. SB acquired funding and supervised the project. LJ and SB contributed to manuscript preparation. Both authors read and approved the final manuscript. Acknowledgements We thank Adrien Leger, Oxford Nanopore Technologies Ltd, for critical reading of the manuscript. We thank Josephine Ilott and Morgan Cockrill , Department of Life Sciences, University of Bath, for their contributions to the NAP pipeline: development of the decontamination module and improvement of the pipeline’s error reporting and overall robustness, respectively. Both contributions have been documented in the project repository. References Adams, A.K. et al. (2023) “Qmatey: an automated pipeline for fast exact matching-based alignment and strain-level taxonomic binning and profiling of metagenomes,” Briefings in Bioinformatics , 24(6), p. bbad351. Available at: https://doi.org/10.1093/bib/bbad351. Arvanitis, M. and Mylonakis, E. (2015) “Fungal–bacterial interactions and their relevance in health,” Cellular Microbiology , 17(10), pp. 1442–1446. Available at: https://doi.org/10.1111/cmi.12493. Camacho, C. et al. (2009) “BLAST+: architecture and applications,” BMC Bioinformatics , 10(1), p. 421. Available at: https://doi.org/10.1186/1471-2105-10-421. Charalampous, T. et al. (2019) “Nanopore metagenomics enables rapid clinical diagnosis of bacterial lower respiratory infection,” Nature Biotechnology , 37(7), pp. 783–792. Available at: https://doi.org/10.1038/s41587-019-0156-5. De Coster, W. et al. (2018) “NanoPack: visualizing and processing long-read sequencing data,” Bioinformatics , 34(15), pp. 2666–2669. Available at: https://doi.org/10.1093/bioinformatics/bty149. Frey-Klett, P. et al. (2011) “Bacterial-Fungal Interactions: Hyphens between Agricultural, Clinical, Environmental, and Food Microbiologists,” Microbiology and Molecular Biology Reviews , 75(4), pp. 583–609. Available at: https://doi.org/10.1128/mmbr.00020-11. Glassing, A. et al. (2016) “Inherent bacterial DNA contamination of extraction and sequencing reagents may affect interpretation of microbiota in low bacterial biomass samples,” Gut Pathogens , 8(1), p. 24. Available at: https://doi.org/10.1186/s13099-016-0103-7. Janda, J.M. and Abbott, S.L. (2007) “16S rRNA Gene Sequencing for Bacterial Identification in the Diagnostic Laboratory: Pluses, Perils, and Pitfalls,” Journal of Clinical Microbiology , 45(9), pp. 2761–2764. Available at: https://doi.org/10.1128/JCM.01228-07. Klindworth, A. et al. (2013) “Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies,” Nucleic Acids Research , 41(1), p. e1. Available at: https://doi.org/10.1093/nar/gks808. Lao, H.-Y. et al. (2021) “The clinical utility of two high-throughput 16S rRNA gene sequencing workflows for taxonomic assignment of unidentifiable bacterial pathogens in MALDI-TOF MS.” bioRxiv, p. 2021.08.16.456588. Available at: https://doi.org/10.1101/2021.08.16.456588. Li, E. et al. (2018) “Benefits of antifungal therapy in asthma patients with airway mycosis: A retrospective cohort analysis,” Immunity, Inflammation and Disease , 6(2), pp. 264–275. Available at: https://doi.org/10.1002/iid3.215. Li, W. and Godzik, A. (2006) “Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences,” Bioinformatics , 22(13), pp. 1658–1659. Available at: https://doi.org/10.1093/bioinformatics/btl158. McNichol, J. et al. (2021) “Evaluating and Improving Small Subunit rRNA PCR Primer Coverage for Bacteria, Archaea, and Eukaryotes Using Metagenomes from Global Ocean Surveys,” mSystems , 6(3), pp. e00565-21. Available at: https://doi.org/10.1128/mSystems.00565-21. “nanoporetech/dorado” (2025). Oxford Nanopore Technologies. Available at: https://github.com/nanoporetech/dorado (Accessed: April 24, 2025). O’Sullivan, D.M. et al. (2021) “An inter-laboratory study to investigate the impact of the bioinformatics component on microbiome analysis using mock communities,” Scientific Reports , 11(1), p. 10590. Available at: https://doi.org/10.1038/s41598-021-89881-2. Palkova, L. et al. (2021) “Evaluation of 16S rRNA primer sets for characterisation of microbiota in paediatric patients with autism spectrum disorder,” Scientific Reports, 11, p. 6781. Available at: https://doi.org/10.1038/s41598-021-86378-w. Parada, A.E., Needham, D.M. and Fuhrman, J.A. (2016) “Every base matters: assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples,” Environmental Microbiology , 18(5), pp. 1403–1414. Available at: https://doi.org/10.1111/1462-2920.13023. Poncheewin, W. et al. (2020) “NG-Tax 2.0: A Semantic Framework for High-Throughput Amplicon Analysis,” Frontiers in Genetics , 10. Available at: https://doi.org/10.3389/fgene.2019.01366. Qin, Y. et al. (2023) “Effects of error, chimera, bias, and GC content on the accuracy of amplicon sequencing,” mSystems , 8(6), pp. e01025-23. Available at: https://doi.org/10.1128/msystems.01025-23. Ramiro-Garcia, J. et al. (2018) “NG-Tax, a highly accurate and validated pipeline for analysis of 16S rRNA amplicons from complex biomes.” F1000Research. Available at: https://doi.org/10.12688/f1000research.9227.2. Rognes, T. et al. (2016) “VSEARCH: a versatile open source tool for metagenomics,” PeerJ , 4, p. e2584. Available at: https://doi.org/10.7717/peerj.2584. Salter, S.J. et al. (2014) “Reagent and laboratory contamination can critically impact sequence-based microbiome analyses,” BMC Biology , 12(1), p. 87. Available at: https://doi.org/10.1186/s12915-014-0087-z. Shen, W. et al. (2016) “SeqKit: A Cross-Platform and Ultrafast Toolkit for FASTA/Q File Manipulation,” PLOS ONE , 11(10), p. e0163962. Available at: https://doi.org/10.1371/journal.pone.0163962. Silverman, J.D. et al. (2021) “Measuring and mitigating PCR bias in microbiota datasets,” PLOS Computational Biology , 17(7), p. e1009113. Available at: https://doi.org/10.1371/journal.pcbi.1009113. Wang, H. et al. (2023) “Intestinal fungi and systemic autoimmune diseases,” Autoimmunity Reviews , 22(2), p. 103234. Available at: https://doi.org/10.1016/j.autrev.2022.103234. Yeh, Y.-C. et al. (2017) “Taxon disappearance from microbiome analysis indicates need for mock communities as a standard in every sequencing run.” bioRxiv, p. 206219. Available at: https://doi.org/10.1101/206219. Yeh, Y.-C. et al. (2021) “Comprehensive single-PCR 16S and 18S rRNA community analysis validated with mock communities, and estimation of sequencing bias against 18S,” Environmental Microbiology , 23(6), pp. 3240–3250. Available at: https://doi.org/10.1111/1462-2920.15553. Youngblut, N.D. et al. (2021) “Vertebrate host phylogeny influences gut archaeal diversity,” Nature Microbiology , 6(11), pp. 1443–1454. Available at: https://doi.org/10.1038/s41564-021-00980-2. Zhang, B. et al. (2020) “Impact of bead-beating intensity on microbiome recovery in mouse and human stool: Optimization of DNA extraction.” bioRxiv, p. 2020.06.15.151753. Available at: https://doi.org/10.1101/2020.06.15.151753. Zhao, J., Rodriguez, J. and Martens-Habbena, W. (2023) “Fine-scale evaluation of two standard 16S rRNA gene amplicon primer pairs for analysis of total prokaryotes and archaeal nitrifiers in differently managed soils,” Frontiers in Microbiology , 14. Available at: https://doi.org/10.3389/fmicb.2023.1140487. ZymoBIOMICS DNA/RNA Miniprep Kit (2025) Zymo Research International . Available at: https://zymoresearch.eu/products/zymobiomics-dna-rna-miniprep-kit (Accessed: April 24, 2025). Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 08 Apr, 2026 Reviews received at journal 02 Apr, 2026 Reviews received at journal 25 Mar, 2026 Reviewers agreed at journal 19 Mar, 2026 Reviewers agreed at journal 16 Mar, 2026 Reviewers invited by journal 04 Mar, 2026 Editor invited by journal 03 Mar, 2026 Editor assigned by journal 26 Nov, 2025 Submission checks completed at journal 26 Nov, 2025 First submitted to journal 21 Nov, 2025 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-8173315\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":false,\"archivedVersions\":[],\"articleType\":\"software\",\"associatedPublications\":[],\"authors\":[{\"id\":601002061,\"identity\":\"15c4ce52-b9c4-4509-90a6-c62e41afad55\",\"order_by\":0,\"name\":\"Luke B. Jones\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"University of Bath\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Luke\",\"middleName\":\"B.\",\"lastName\":\"Jones\",\"suffix\":\"\"},{\"id\":601002063,\"identity\":\"8b930877-ac14-498d-8151-1b668b4876b1\",\"order_by\":1,\"name\":\"Stefan Bagby\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuUlEQVRIiWNgGAWjYBACAyBmZqg4wMAggRAgRsuZAww8pGlhbCNFizn/8cefC+fdkbeXbmD88IPhsDFBLZYzcsykZ257Ztgjc4BZsofhsBlhh93gYWPm3XaYsUcigUGageGwDWEt54EO451z2B6ohfk3cVoOJBhI8zYcTgRqYQPZQthhYL/MOPYsuedGYptlj0E6Ye+DQ6yg5o5t+4zkwzd+VFgbNhDUgwCMDcTEyigYBaNgFIwCYgAAUXQ6zfofVqEAAAAASUVORK5CYII=\",\"orcid\":\"\",\"institution\":\"University of Bath\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Stefan\",\"middleName\":\"\",\"lastName\":\"Bagby\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2025-11-21 11:53:21\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-8173315/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-8173315/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":104321356,\"identity\":\"69807432-661a-453f-8fff-5f6784a30579\",\"added_by\":\"auto\",\"created_at\":\"2026-03-10 13:19:57\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":133997,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eCore Pipeline Workflow. (A) Dorado output is the input to the pipeline, and is fed into the wrapper nap.sh which passes variables along with the config to pipe.sh, initiating the pipeline. (B) Quality control stages shown in light blue. (C) ‘RAW’ data are clustered and the output centroids are then taxonomically classified, producing the ‘CNT’ dataset. RAW data are then BLASTed against CNT data, adding the abundance dimension to the high confidence taxonomy hits. (D) Final BLAST outputs are simplified to species level, collapsing taxonomy, and combining abundances accordingly. This dataset is then normalised using total sum calling (TSS) normalisation. If blanks are set up correctly (using ‘nap decon’), Decontamination.py then removes organisms based on blank content, outputting Microbiome.tsv. Tools used: NanoFilt (v2.8.0; De Coster \\u003cem\\u003eet al.,\\u003c/em\\u003e2018), SeqTK (v1.4; Shen \\u003cem\\u003eet al\\u003c/em\\u003e., 2016), VSEARCH (v2.28.1; Rognes \\u003cem\\u003eet al\\u003c/em\\u003e., 2016), BLAST (v2.16.0; Camacho \\u003cem\\u003eet al\\u003c/em\\u003e., 2009), and CD-HIT (4.8.1; Li and Godzik, 2006).\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/cc423c7767b5132c46807412.png\"},{\"id\":104405483,\"identity\":\"e4b9a4d8-46c2-4229-80be-5948c5da8070\",\"added_by\":\"auto\",\"created_at\":\"2026-03-11 12:23:06\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":82448,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003ePhred to Read Depth and CNT to RAW Read Depth Scatterplots. (A) Filtered read count (‘RAW’ count used to find centroids and apply abundances) against the input count. (B) Fraction of reads lost to filtration, plotted against the Phred threshold applied during filtration. L1-L3 are replicates of the logarithmic mock community, and M1-M3 are replicates of the gut mock community.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/a4df40b9875dacfd70954885.png\"},{\"id\":104321354,\"identity\":\"232a2561-f53f-4bbb-8b2b-608bea6dd8b5\",\"added_by\":\"auto\",\"created_at\":\"2026-03-10 13:19:57\",\"extension\":\"png\",\"order_by\":3,\"title\":\"Figure 3\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":81362,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eβ-diversity of Pipeline Refinement and Final Results Relative to Theoretical Sample Content at Genus-Level. Grey points show data points retrieved from pipeline refinement stages, coloured data points identify expected content (X) and observed outputs of the final pipeline (dots). The inserts show specifically the final data points, and are annotated with b-diversity metrics for proximity of an output to its corresponding hypothetical point. L1-L3 are replicates of the logarithmic mock community, and M1-M3 are replicates of the gut mock community.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/55b3df3e5a9c99707a4341e6.png\"},{\"id\":104321357,\"identity\":\"e4ebc47b-4259-4af8-b3da-937df6ec60d1\",\"added_by\":\"auto\",\"created_at\":\"2026-03-10 13:19:57\",\"extension\":\"png\",\"order_by\":4,\"title\":\"Figure 4\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":167570,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eObserved Taxonomic Abundances Across Mock Communities at Genus and Species Levels. (A) Genus level results. (B) Species level results. Subpanels (1) show the gut microbiome mock (M1–M3) and subpanels (2) show the logarithmic mock (L1–L3). Grey bars represent expected abundances, while coloured points indicate observed abundances\\u003cem\\u003e per \\u003c/em\\u003ereplicate. Coefficient of variation for expected taxa across repeats is annotated above bars; values \\u0026gt;0.5 are coloured red. The panels are annotated with Pearson correlation coefficient (\\u003cem\\u003er\\u003c/em\\u003e) and p-value (\\u003cem\\u003ep\\u003c/em\\u003e) between mean observed and expected abundances. All detected taxa are included in plots and statistical analyses, including contaminants and false positives.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"4.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/21d8c8dc09009b31af9a2999.png\"},{\"id\":104321359,\"identity\":\"f074e2c2-671e-42c2-800c-a1db45447ccb\",\"added_by\":\"auto\",\"created_at\":\"2026-03-10 13:19:57\",\"extension\":\"png\",\"order_by\":5,\"title\":\"Figure 5\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":192137,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eAgreement and Error in Taxonomic Quantification Across Mock Communities at Genus Level.\\u003cem\\u003e\\u003cbr\\u003e\\n(A) Gut mock community (M1–M3). (B) Logarithmic mock community (L1–L3). Left-hand plots show agreement between expected and observed per taxon genus level abundances. Agreement plots are annotated with Pearson correlation coefficient (r), p-value (p), and Lin’s concordance correlation coefficient (ρc). Right-hand plots show Bland–Altman analysis, revealing mean bias and spread of differences across replicates, with bias annotated. This analysis (both agreement and Bland-Altman) only includes taxa present in the expected profile and their corresponding observed abundances; false positives are excluded.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"5.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/99fed486736edf31eb06082c.png\"},{\"id\":104321360,\"identity\":\"55688131-26d8-4163-9366-ae6d2c370132\",\"added_by\":\"auto\",\"created_at\":\"2026-03-10 13:19:57\",\"extension\":\"png\",\"order_by\":6,\"title\":\"Figure 6\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":188956,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003eA\\u003c/em\\u003egreement and Error in Taxonomic Quantification Across Mock Communities at Species Level.\\u003cem\\u003e\\u003cbr\\u003e\\n(A) Gut mock community (M1–M3). (B) Logarithmic mock community (L1–L3). Left-hand plots show agreement between expected and observed per-taxon species level abundances. Agreement plots are annotated with Pearson correlation coefficient (r), p-value (p), and Lin’s concordance correlation coefficient (ρc). Right-hand plots show Bland–Altman analysis, revealing mean bias and spread of differences across replicates, with bias annotated. This analysis (both agreement and Bland-Altman) only includes taxa present in the expected profile and their corresponding observed abundances; false positives are excluded.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"6.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/a997faec79e2c62608f9b109.png\"},{\"id\":104405523,\"identity\":\"25527838-057e-4095-b971-38533b81fdc5\",\"added_by\":\"auto\",\"created_at\":\"2026-03-11 12:23:11\",\"extension\":\"png\",\"order_by\":7,\"title\":\"Figure 7\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":119286,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eDetection Sensitivity and Classification Performance Across Mock Communities. (A) \\u003cem\\u003eLogarithmic mock community. \\u003c/em\\u003e(B)\\u003cem\\u003eGut mock community. \\u003c/em\\u003eLeft:detection sensitivity across expected taxa. Each point represents a taxon, plotted by its expected relative abundance (%) and the fraction of replicates in which it was detected. Right:precision–recall scatter \\u003cem\\u003eper\\u003c/em\\u003e replicate (threshold τ ≥ 0.001). True positive, false positive, false negative, and true negative rates were calculated to define the precision-recall metrics.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"7.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/6ac345a6144aa9a1db17a556.png\"},{\"id\":104409395,\"identity\":\"27a59d51-f28c-4389-8ec3-1791efc9e449\",\"added_by\":\"auto\",\"created_at\":\"2026-03-11 12:45:00\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":1536167,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8173315/v1/f29e40a9-e865-4496-b09e-4672955be5b3.pdf\"}],\"financialInterests\":\"No competing interests reported.\",\"formattedTitle\":\"NAP: An Open-Source Pipeline for Cross-Domain Microbiome Profiling Using Nanopore Sequencing-Derived Amplicon Data\",\"fulltext\":[{\"header\":\"Background\",\"content\":\"\\u003cp\\u003eMicrobiomes are increasingly studied from a cross-domain perspective, capturing interactions among bacteria, archaea, and eukaryotes across diverse environmental and host-associated contexts. While microbiome research has historically focused on bacteria, recent work highlights that inter-domain interactions are central to understanding community structure and function, and health-related outcomes. In clinical settings, bacterial–fungal interactions have physical, metabolic, and immune implications in disease (Arvanitis and Mylonakis, \\u003cspan class=\\\"CitationRef\\\"\\u003e2015\\u003c/span\\u003e). Inter-kingdom microbial networks are now recognized as potential drivers of complex pathologies and may offer novel diagnostic and therapeutic targets (Li et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2018\\u003c/span\\u003e; Wang et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e). Beyond medicine, these interactions shape ecological dynamics, influence food microbiology, and affect agricultural productivity (Frey-Klett et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2011\\u003c/span\\u003e). As such, accurate and domain-inclusive profiling of microbial communities is essential for advancing microbiome science across multiple disciplines.\\u003c/p\\u003e \\u003cp\\u003eWhile shotgun and whole genome sequencing approaches offer high taxonomic resolution and low taxonomic bias, they remain costly and computationally intensive. For studies focused primarily on community composition, amplicon sequencing provides a faster and more cost-effective alternative. Traditionally, however, amplicon-based profiling has been restricted to bacteria using 16S rRNA primers, or has relied on separate 16S and 18S reactions to study bacterial and eukaryotic components independently. Because these are amplified and sequenced in isolation, their abundance outputs are not directly comparable, limiting cross-domain integration.\\u003c/p\\u003e \\u003cp\\u003eRecent advances in primer design have addressed this issue by enabling the amplification of bacterial, archaeal, and eukaryotic small subunit rRNA from a single reaction using universal primer sets. Illumina-based workflows offer high base accuracy and are compatible with such primers, but remain limited by short read lengths. Moreover, while mature and user-friendly pipelines exist for Illumina data, these often require programming expertise and lack tools for domain-aware normalisation or bias correction. Nanopore sequencing, in contrast, supports real-time, full length amplicon sequencing with much lower capital outlay and greater portability (Lao et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2021\\u003c/span\\u003e; Charalampous et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2019\\u003c/span\\u003e). These features make it an appealing choice for field-based or resource-limited settings. However, existing Nanopore amplicon tools are largely limited to bacterial 16S pipelines (e.g., EPI2ME) or require researchers to build bespoke workflows from scratch, rendering them inaccessible to most potential users, particularly those aiming to study mixed-domain microbial communities.\\u003c/p\\u003e \\u003cp\\u003eWe present here the Nanopore sequencing-based Amplicon Pipeline (NAP), a lightweight, open-source pipeline purpose built to transform Nanopore amplicon sequencing data into high quality, bias-corrected, and taxonomically normalised microbiome profiles. Unlike existing pipelines which are either tailored exclusively for bacterial profiling or require extensive computational expertise, NAP is designed to accommodate any cross-domain rRNA primer sets (such as the default set, 515Y/926R) supporting simultaneous bacterial, archaeal, and eukaryotic profiling from a single amplicon pool.\\u003c/p\\u003e\\n\\u003ch3\\u003eImplementation\\u003c/h3\\u003e\\n\\u003cp\\u003eSILVA rRNA reference databases are pre-filtered to exclude ambiguous, unclassified, or contaminant entries (e.g., plant mitochondrial rRNA), and are split into 16S and 18S subsets. Primer sequences are then mapped to each entry, extracting the longest amplicon region \\u003cem\\u003eper\\u003c/em\\u003e entry \\u003cem\\u003evia\\u003c/em\\u003e ungapped local alignment; untrimmed entries are retained where necessary. These domain-specific trimmed databases are then indexed for BLAST-based classification.\\u003c/p\\u003e \\u003cp\\u003eNAP processes demultiplexed amplicon reads (FASTQ) through a modular workflow optimised for mixed-domain microbiome profiling. Reads are first filtered dynamically, where read depth and quality are optimised by adjusting quality thresholds and masking low quality bases. Chimeric reads are then removed, and the remaining reads (termed the RAW dataset) are clustered into centroids (termed CNT dataset; Fig.\\u0026nbsp;\\u003cspan class=\\\"InternalRef\\\"\\u003e1\\u003c/span\\u003eA and B). Centroid sequences are classified \\u003cem\\u003evia\\u003c/em\\u003e BLAST against the custom SILVA reference database, and taxonomic identities are assigned using a majority consensus algorithm. These annotated CNT sequences then serve as a high-confidence internal database to reclassify all RAW reads against, boosting precision while reducing computational load (Fig.\\u0026nbsp;\\u003cspan class=\\\"InternalRef\\\"\\u003e1\\u003c/span\\u003eC). Abundance data are normalised using total sum scaling (TSS), corrected for known 16S/18S domain bias, and filtered to remove contaminants and low confidence taxa. Final outputs include genus- and species-level abundance tables (.tsv), pipeline logs (.txt), and quality summaries to guide user confidence (Fig.\\u0026nbsp;\\u003cspan class=\\\"InternalRef\\\"\\u003e1\\u003c/span\\u003eD).\\u003c/p\\u003e \\u003cp\\u003e \\u003c/p\\u003e \\u003cp\\u003eNAP is implemented in Bash and Python and is designed for Unix/Linux-based environments. It is operated entirely \\u003cem\\u003evia\\u003c/em\\u003e the command line, with no programming required beyond Conda and installation. Users configure the pipeline \\u003cem\\u003evia\\u003c/em\\u003e a main configuration file that sets key parameters such as quality thresholds, filtering behaviour, and reference paths. Primer-specific settings are handled \\u003cem\\u003evia\\u003c/em\\u003e modular sub-configuration files, allowing users to easily adapt the workflow to different primer sets, experimental designs, and hardware usage without altering the core code. Installation instructions are available \\u003cem\\u003evia\\u003c/em\\u003e GitHub (\\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://github.com/Luke-B-Jones/NAP\\u003c/span\\u003e\\u003cspan class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003eThe pipeline demonstrates robust genus- and species-level classification accuracy, handles low depth and high noise data gracefully, and ensures reproducible outputs through consistent detection and strong agreement metrics. NAP thus fills a niche between highly accurate but capital- and infrastructure-intensive Illumina approaches and field-compatible emerging Nanopore workflows. By combining adaptability, accessibility, and analytical rigour, it unlocks the potential of Nanopore sequencing for rapid, accurate, and cross-domain microbiome analysis across contexts.\\u003c/p\\u003e \\u003cdiv id=\\\"Sec3\\\" class=\\\"Section2\\\"\\u003e \\u003cdiv id=\\\"Sec4\\\" class=\\\"Section3\\\"\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \"},{\"header\":\"Methods\",\"content\":\"\\u003ch2\\u003eMock Community Preparation\\u003c/h2\\u003e\\u003cp\\u003eTwo commercially available mock microbiome communities were used to validate the pipeline: ZymoBIOMICS Microbial Community Standard II (Log Distribution), and ZymoBIOMICS Gut Microbiome Standard (both Cambridge Biosciences); henceforth, these are referred to as log mock (L1, L2, and L3) community and gut mock (M1, M2, and M3) community, respectively. In addition to this, environmental blanks were made by exposing sterile 0.9% saline to the laboratory environment. DNA was extracted from these samples using ZymoBIOMICS DNA/RNA miniprep kit according to the manufacturer’s recommendations (“ZymoBIOMICS DNA/RNA Miniprep Kit,” 2025). Homogenisation was done according to a custom protocol, with six cycles of 30 seconds at 9,000 rpm using a Bertin Precellys Evolution bead beater, increasing the yield of harder-to-lyse organisms (Zhang et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2020\\u003c/span\\u003e). DNA was quantified using a Qubit fluorometer (Thermo Fisher Scientific). Isolates were then amplified using Phusion™ Plus PCR Master Mixes (Thermo Fisher Scientific), using an annealing temperature of 50°C and \\u003cem\\u003eca.\\u003c/em\\u003e 100ng of template DNA \\u003cem\\u003eper\\u003c/em\\u003e reaction, along with the recommended guidelines for 50 µL single reaction protocol outlined by the supplier, and custom oligonucleotide primers (Eurofins) (515Y, 5′-\\u003cem\\u003eGTGYCAGCMGCCGCGGTAA\\u003c/em\\u003e, 926R, 5′-\\u003cem\\u003eCCGYCAATTYMTTTRAGTTT;\\u003c/em\\u003e McNichol et al., \\u003cspan class=\\\"CitationRef\\\"\\u003e2021\\u003c/span\\u003e). All amplifications were validated using gel electrophoresis by identifying 16S and 18S bands at \\u003cem\\u003eca.\\u003c/em\\u003e 300–500 and \\u003cem\\u003eca.\\u003c/em\\u003e 700 bases, respectively. This resulted in six amplicon pools, three replicates \\u003cem\\u003eper\\u003c/em\\u003e mock community. Amplicon pools were then sequenced using one MinION R10.4.1 flow cell according to the Native Barcoding Protocol (SQK-NBD114.24), and then basecalled and demultiplexed using Dorado (v0.9.0; “nanoporetech/dorado,” 2025; super accuracy model v5.0.0).\\u003c/p\\u003e\\u003ch2\\u003eStatistical Analysis\\u003c/h2\\u003e\\u003cp\\u003eStatistical analysis was performed using a custom Python script to evaluate the accuracy, reliability, and structural validity of the pipeline’s outputs relative to the known composition of two commercially available mock communities. Genus- and species-level taxonomic tables were generated from each sample. To enable comprehensive sensitivity profiling, taxa absent across all replicates were assigned zero counts.\\u003c/p\\u003e\\u003cp\\u003eDetection accuracy was quantified by calculating precision, recall, and F1-scores for each replicate at a fixed detection threshold (τ = 0.001), using the expected genera as the reference set. To evaluate quantitative agreement between observed and expected taxon abundances, linear regression and Lin’s concordance correlation coefficient (\\u003cem\\u003eρ\\u003c/em\\u003e\\u003csub\\u003eCCC\\u003c/sub\\u003e) were used. This enabled both proportional and absolute agreement to be assessed. Agreement plots compared mean observed abundances across replicates against their expected values, while Bland–Altman plots quantified systematic bias and calculated limits of agreement using ± 1.96 standard deviations from the mean difference. Together, these analyses assessed not only detection but also abundance accuracy and consistency. This was restricted to taxa which were expected, meaning false positives were not included in this part of the analysis.\\u003c/p\\u003e\\u003cp\\u003eCommunity-level similarity and sample structure were explored using β-diversity metrics. Pairwise dissimilarities were calculated using both Jaccard distance (based on presence/absence data) and Bray–Curtis dissimilarity (based on relative abundances). These dissimilarity matrices were visualised using principal coordinate analysis (PCoA) to identify clustering of replicates around their respective mock community centroids. A replicate was defined as a success if it was closer in β-diversity space to its own mock profile than to that of the alternate mock group. Statistical significance of replicate fidelity was assessed using one-sided binomial tests for each group, with groupwise \\u003cem\\u003ep\\u003c/em\\u003e-values aggregated \\u003cem\\u003evia\\u003c/em\\u003e Fisher’s method to test for global consistency. To investigate taxon-level variability, bar plots were generated showing the expected genus abundances overlaid with individual replicate values. For each genus, the coefficient of variation (CV) across replicates was annotated, with values exceeding 0.5 highlighted to indicate high within-group variability. Detection sensitivity was further examined by plotting the proportion of replicates that detected each expected taxon against its relative abundance, providing a measure of detection robustness at different abundance levels. These analyses included all observed taxa, including false positives.\\u003c/p\\u003e\"},{\"header\":\"Results\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003eRead Filtering and Quality Control\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eNAP applies an automated read level filtering step that dynamically adjusts Phred score thresholds to balance retained depth and base quality, thereby optimising downstream taxonomic accuracy. After filtering, the number of retained reads correlates positively with the initial input read count (Figure 2A). \\u0026nbsp; The observed variance around this trend is primarily attributable to sample-specific error profiles arising from variability in sequencing throughput, amplification efficiency, and the initial DNA integrity. To limit computational overhead, samples are deliberately downsampled to 225,000 reads, rather than allowing excessive retention after filtering; this is observed in the logarithmic mock repeats, with read counts being capped at 225,000 (Figure 2A). In most samples, the adaptive filtering maintained thresholds above Q30 (average Phred score of reads) while retaining over 100,000 high quality reads (Figures 2A, 2B). For lower quality samples such as M1, the pipeline adaptively reduced the threshold to Q24, masking bases below Q5 (corresponding to an error probability of ~31.6%, and representing less than 0.01% of all bases) rather than discarding more reads entirely. This preserved sufficient depth (\\u0026gt;10,000 reads) while minimising erroneous taxonomic assignments. This was done to a lesser extent in higher depth samples; for example, mock L1 was masked below Q1 (corresponding to an error probability of ~79.4%, and affecting fewer than 0.001% of all bases). Chimera removal using VSEARCH led to the exclusion of 8.55% \\u0026plusmn; 3.37% of reads on average, meaning read depths were not substantially affected downstream. Overall, NAP\\u0026rsquo;s default filtering effectively balanced read depth and quality, though user-defined parameter tuning may improve performance for datasets with atypical quality profiles.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003ePipeline Refinement\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eInitial taxonomic assignments were refined to improve community structure fidelity and mitigate Nanopore-specific sequencing errors, including substitution and indel artefacts. Multiple refinement strategies were evaluated for recovering accurate amplicon sequence variants (ASVs). Methods included clustering-based approaches such as CD-HIT, isONclust, and Rattle to identify sequence centroids. In parallel, alignment-based consensus calling with Medaka, Racon, and transcript-style workflows were evaluated, using both reference-guided and centroid-guided strategies. Taxonomic classification methods were compared using Kraken2, QIIME2, and BLAST to assess accuracy and robustness across toolkits. Consensus sequence\\u0026ndash;polishing approaches (e.g., Medaka, Racon, and custom tools) were inconsistent, particularly for low abundance taxa, and frequently propagated sequencing errors into final profiles. In contrast, BLAST provided tuneable alignment parameters that could be optimised for Nanopore-specific error profiles. Our BLAST with CD-HIT configuration improved centroid-level taxonomic assignment accuracy relative to other methods. BLAST also enabled a post hoc consensus voting algorithm to efficiently resolve classification ambiguities at the species level. The speed of pipeline completion was improved by SILVA database refinement (trimming regions not amplified by primer set, and removing taxonomically unresolved entries e.g., uncultured, metagenomically assembled), and this increased consensus algorithm accuracy (partially due to refined regions, but also by removing ambiguous entries). Using this approach, classification rates ranged from 98.53% to 99.88% in \\u0026gt;Q30 samples, and reached 94.68% in Q20 samples (species level); this is comparable to the performance of NG-Tax, an Illumina-based amplicon pipeline (Ramiro-Garcia \\u003cem\\u003eet al\\u003c/em\\u003e., 2018), under tested conditions.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u0026beta;-diversity analyses showed strong concordance between observed and expected profiles at both genus and species levels. For all mock communities, genus level Bray\\u0026ndash;Curtis and Jaccard distances were \\u0026le; 0.5 (Figure 3). At the species level, the logarithmic mock community achieved Bray\\u0026ndash;Curtis distances ranging from 0.01 to 0.13 and Jaccard distances from 0.33 to 0.50. The gut mock community exhibited slightly higher but consistent Bray\\u0026ndash;Curtis distances (0.44\\u0026ndash;0.47) and Jaccard distances (0.33\\u0026ndash;0.45). In contrast, alternative refinement strategies tested during pipeline development frequently exceeded 0.6 for both metrics. These results support the final pipeline configuration, centred on CD-HIT clustering and BLAST-based classification with consensus correction, as the most accurate and reproducible method. The clear convergence toward expected taxonomic structure supports its effectiveness for Nanopore-based microbiome analysis (Figure 3).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eTaxonomic Classification\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eAll expected bacterial and eukaryotic genera were detected across both mock communities, with only Saccharomyces exhibiting high replicate variability (coefficient of variation \\u0026gt; 0.5; Figure 4A). The only archaeal genus (Methanobrevibacter) \\u0026nbsp;was detected below the pipeline\\u0026rsquo;s 0.05% confidence threshold. Similarly, Salmonella was detected but also fell below the applied filtering threshold. Only four genera were identified as false positives at the genus level: Shigella, likely reflecting misclassification due to the presence of multiple Escherichia strains and their known sequence similarity within the V4\\u0026ndash;V5 small subunit ribosomal RNA regions; and Bysmatrum, Pasteurellaceae, and Streptococcus, which displayed sporadic, highly variable abundances and were also present in blank controls, suggesting contamination beyond the detection limits of the current decontamination method. This interpretation is supported by previous reports identifying Lactobacillus, Saccharomyces, and Streptococcus as common laboratory- or human-associated contaminants (Glassing et al., 2016; Salter et al., 2014).\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u0026nbsp;Taxonomic accuracy was mostly preserved at the species level. However, \\u003cem\\u003eVeillonella rogosae was not detected in any gut mock replicates; instead, artefactual misclassification occurred under V. dispar and V. parvus. A similar but less pronounced pattern was observed for Lactobacillus fermentum, which showed inflated abundance due to misattribution to L. acidophilus, a known and present contaminant. These cases are consistent with the high sequence similarity of the V4\\u0026ndash;V5 region among closely related taxa (Janda and Abbott, 2007). For both Veillonella and Lactobacillus, minor Nanopore sequencing errors likely contributed to taxonomic divergence during species level classification.\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eAt both genus and species levels, replicate profiles were statistically similar to the expected mock community composition (p \\u0026lt; 0.05; Figure 4), indicating high reproducibility and classification accuracy. Most expected taxa were successfully identified with low inter-replicate variability (e.g., CV \\u0026lt; 0.4 in most cases, often reaching \\u0026lt; 0.15). Contaminants were identified based on prevalence in blank controls, high blank-to-sample abundance ratios, and exclusion from expected taxa lists. On average, the decontamination process removed 7.00 \\u0026plusmn; 2.68 species level hits per replicate, all of which were present in blank samples and judged to be contaminants following manual review. An additional 9.83 \\u0026plusmn; 6.49 species per replicate exhibited altered abundances after decontamination without being fully removed. For several high abundance taxa (such as Faecalibacterium prausnitzii, Listeria monocytogenes, and Pseudomonas aeruginosa), these adjustments moved observed abundances closer to expected values and reduced variability across replicates (Figure 4). Together, t\\u003c/em\\u003ehese findings indicate that, although the current decontamination strategy is relatively simple and does not fully resolve artefactual misclassification, it effectively identifies common contaminants and improves taxonomic accuracy. Classification performance is highly reliable at the genus level, and with improved primer design or longer read lengths, species level accuracy could be enhanced.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eTaxon Abundance Accuracy\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe pipeline\\u0026rsquo;s default primers, 515Y/926R, target the V4\\u0026ndash;V5 region of the small subunit (SSU) rRNA and offer broad cross-domain coverage, amplifying \\u003cem\\u003eca.\\u0026nbsp;\\u003c/em\\u003e96% of known rRNA sequences across bacteria, archaea, and eukaryotes. Although these primers improve upon earlier cross-domain designs, they still introduce bias through differential binding site specificity and annealing efficiency, which can skew taxon representation (Parada \\u003cem\\u003eet al\\u003c/em\\u003e., 2016; McNichol \\u003cem\\u003eet al\\u003c/em\\u003e., 2021). Such effects are further compounded by wet lab variables, particularly DNA extraction efficiency, and SSU copy number variation. To mitigate these biases, we implemented prolonged bead beating cycles to balance DNA recovery from both easily lysed Gram-negative bacteria and more resilient taxa, including Gram-positive bacteria, spore formers, and eukaryotes (Zhang \\u003cem\\u003eet al\\u003c/em\\u003e., 2020).\\u003c/p\\u003e\\n\\u003cp\\u003eWe assessed the accuracy of relative abundance estimates by comparing observed taxon abundances against expected values from mock community profiles using agreement plots and Bland\\u0026ndash;Altman analysis at genus level (Figure 5). The logarithmic mock community showed excellent concordance, with minimal Bland\\u0026ndash;Altman bias (21.2 units) and strong statistical agreement across replicates (\\u003cem\\u003er\\u003c/em\\u003e = 1, \\u003cem\\u003ep\\u003c/em\\u003e = 1.5e\\u003csup\\u003e-5\\u003c/sup\\u003e, \\u003cem\\u003ep\\u003c/em\\u003e\\u003csub\\u003eCCC\\u003c/sub\\u003e = 1). In contrast, the gut mock community showed reduced correlation (\\u003cem\\u003er\\u003c/em\\u003e = 0.68, \\u003cem\\u003ep\\u003c/em\\u003e = 2.5e\\u003csup\\u003e-3\\u003c/sup\\u003e, \\u003cem\\u003ep\\u003c/em\\u003e\\u003csub\\u003eCCC\\u003c/sub\\u003e = 0.53) and a substantial negative bias (\\u0026ndash;5876.6), suggesting systematic underrepresentation of several expected taxa. However, intra-taxon variability remained low (Figure 4A), indicating that discrepancies were driven by consistent biases, potentially due to primer mismatch or extraction inefficiencies, rather than random noise or computational error. To further validate the pipeline\\u0026rsquo;s reliability, we repeated the agreement and Bland\\u0026ndash;Altman analyses at species level (Figure 6). Results remained statistically robust, with only marginal reductions in correlation (e.g., gut mock: \\u003cem\\u003er\\u003c/em\\u003e = 0.68 genus vs. 0.66 species; \\u003cem\\u003ep\\u003c/em\\u003e = 2.5\\u0026times;10⁻\\u0026sup3; vs. 1.7\\u0026times;10⁻\\u0026sup3;; \\u003cem\\u003e\\u0026rho;\\u003csub\\u003eCCC\\u003c/sub\\u003e\\u003c/em\\u003e = 0.53 vs. 0.56) and reduced Bland\\u0026ndash;Altman bias in both mock communities (e.g., log mock: 21.2 genus vs. 17.7 species). These patterns suggest that, in well characterised taxa, species level profiles produced by the pipeline remain quantitatively trustworthy. Nonetheless, while species level abundances appear numerically more concordant, taxonomic resolution is often compromised by limitations of the V4\\u0026ndash;V5 region, which can fragment expected taxa into multiple spurious species (e.g., \\u003cem\\u003eVeillonella\\u003c/em\\u003e being misassigned to two absent species). As such, genus level assignments remain more biologically reliable according to these results, and are therefore preferred for robust interpretation under the pipeline\\u0026rsquo;s default settings.\\u003c/p\\u003e\\n\\u003cp\\u003eWhile 515Y/926R taxonomic bias has not been well explored, there is evidence that the observed abundance discrepancies are not attributable to pipeline error. The phylum \\u003cem\\u003ePseudomonadota\\u003c/em\\u003e (formerly\\u0026nbsp;\\u003cem\\u003eProteobacteria\\u003c/em\\u003e) was well represented, with\\u0026nbsp;\\u003cem\\u003ePseudomonas\\u003c/em\\u003e showing expected abundances, as observed in other studies (Klindworth \\u003cem\\u003eet al\\u003c/em\\u003e., 2013). Conversely, as expected,\\u0026nbsp;\\u003cem\\u003eActinobacteriota\\u003c/em\\u003e was underrepresented, with\\u0026nbsp;\\u003cem\\u003eBifidobacterium\\u003c/em\\u003e displaying particularly low abundance. Amplicon bias is known in this phylum, and is potentially compounded here by the Gram-positive cell wall structure (Parada \\u003cem\\u003eet al.\\u003c/em\\u003e, 2016). Bacillota (previously\\u0026nbsp;\\u003cem\\u003eFirmicutes)\\u003c/em\\u003e were moderately underrepresented overall, while\\u0026nbsp;\\u003cem\\u003eBacteroidetes\\u003c/em\\u003e were more substantially affected, a pattern of particular interest given the clinical relevance of the \\u0026lsquo;Firmicutes:Bacteroidetes\\u0026rsquo; ratio (Palkova \\u003cem\\u003eet al\\u003c/em\\u003e., 2021). Within these phyla,\\u0026nbsp;\\u003cem\\u003eFaecalibacterium\\u003c/em\\u003e,\\u0026nbsp;\\u003cem\\u003eLactobacillus, Listeria\\u003c/em\\u003e, and\\u0026nbsp;\\u003cem\\u003eVeillonella\\u003c/em\\u003e showed good or mildly reduced recovery, while\\u0026nbsp;\\u003cem\\u003eBacteroides and Prevotella\\u003c/em\\u003e were more strongly underrepresented. This skew was less pronounced than in some prior studies, possibly due to differential lysis efficiency: Gram-positive Bacillota may be more effectively extracted than Gram-negative Bacteroidetes under prolonged bead beating conditions (Parada \\u003cem\\u003eet al.\\u003c/em\\u003e, 2016; Zhao \\u003cem\\u003eet al\\u003c/em\\u003e., 2023).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eBacillus\\u003c/em\\u003e detection was inconsistent across replicates. In two gut mock community samples, it was present below the reporting threshold; in a third, it appeared overrepresented. As a spore forming, Gram-positive taxon,\\u0026nbsp;\\u003cem\\u003eBacillus\\u003c/em\\u003e is known to be sensitive to extraction and amplification biases. Additionally, Table 1 supports the possibility of mismatch-based primer bias, a key factor in under-amplification of rRNA. Furthermore, its presence in extraction blanks suggests potential cross-contamination, meaning its abundances are further reduced. Misalignment may also explain the inflated abundance in one replicate, given the high sequence similarity between\\u0026nbsp;\\u003cem\\u003eBacillus\\u003c/em\\u003e strains, where a single significant contamination event caused reinforcement of the taxon\\u0026rsquo;s presence. A comparable pattern was seen with\\u0026nbsp;\\u003cem\\u003eShigella\\u003c/em\\u003e, which was not present in the mock community but appeared alongside\\u0026nbsp;\\u003cem\\u003eEscherichia\\u003c/em\\u003e, likely due to shared rRNA sequence regions and minor contamination (Zhao \\u003cem\\u003eet al\\u003c/em\\u003e., 2023).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eOther taxa have poorly characterised biases for primer set 515Y/926R, limiting interpretation. Here, for example,\\u0026nbsp;\\u003cem\\u003eMethanobrevibacter smithii and Salmonella enterica were both detected below the pipeline\\u0026rsquo;s filtering threshold (0.05% relative abundance filtration threshold; 0.2% and 0.089% expected relative abundance for M. smithii and S. enterica, respectively), suggesting mild underrepresentation, with only Salmonella enterica showing potential for primer mismatches (Table 1). Similarly, Akkermansia muciniphila showed no primer mismatches and underrepresentation, with bias uncharacterised in the literature. For Methanobrevibacter smithii, the difficulty of archaeon lysis is a strong candidate contributor, but other factors can lead to related archaea being underrepresented\\u003c/em\\u003e (Youngblut \\u003cem\\u003eet al.,\\u003c/em\\u003e 2021; Zhao\\u003cem\\u003e\\u0026nbsp;et al.\\u003c/em\\u003e, 2023).\\u003cem\\u003e\\u0026nbsp;Akkermansia muciniphila and Salmonella enterica are easy to lyse Gram-negative, non-spore forming bacteria, suggesting excessive bead beating may have led to a slight underrepresentation bias, but this is likely compounded by other PCR-related biases, considering mismatches are not the only known factor in primer-induced bias\\u0026nbsp;\\u003c/em\\u003e(Qin \\u003cem\\u003eet al.\\u003c/em\\u003e, 2023; Silverman \\u003cem\\u003eet al\\u003c/em\\u003e., 2021).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eFinally, as the 515Y/926R primers also capture eukaryotic 18S\\u003c/em\\u003e rRNA, a domain correction factor is required to adjust for lower 18S amplification efficiency. The pipeline applies a default correction factor of 0.4, based on empirical estimates ranging from 0.3\\u0026ndash;0.5. This correction was validated in our results: \\u003cem\\u003eCandida albicans\\u003c/em\\u003e and \\u003cem\\u003eSaccharomyces cerevisiae\\u003c/em\\u003e were detected at or near to the expected abundances relative to co-occurring 16S taxa (Yeh \\u003cem\\u003eet al\\u003c/em\\u003e., 2021).\\u003c/p\\u003e\\n\\u003cp\\u003eTaken together, replicate abundances were highly consistent, demonstrating strong technical reproducibility. Deviations from expected values were largely attributable to primer mismatch, variability in lysis efficiency, or low input abundance (rather than stochastic or computational error). The pipeline\\u0026rsquo;s default 16S/18S correction proved effective, and the bead beating protocol did not seem to induce a systematic skew between Gram-positive and Gram-negative bacterial taxa. These findings support the pipeline\\u0026rsquo;s capacity to generate biologically representative taxonomic profiles across complex microbiomes.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eTable 1:\\u0026nbsp;\\u003c/strong\\u003eTestPrime Output Assessing Primer Bias Across Observed Genera. TestPrime was run using SSU SILVA version 138.2 RefNR (pruned to remove redundancy). Mismatches were configured to permit three mismatches, with no enforced 0-mismatch zone at the 3\\u0026prime; end. Results indicated 96.6% coverage of the SILVA database. The forward primer was responsible for mismatches in 50.6% of cases, and the reverse primer in 46.0% of cases, while mismatches occurred in both primers in 3.4% of cases; these results refer to an \\u003cstrong\\u003e\\u003cem\\u003ein silico\\u003c/em\\u003e\\u003c/strong\\u003e evaluation of theoretical primer binding, and therefore do not account for other external factors. The \\u0026quot;Perfect Match\\u0026quot; column shows the percentage of references that aligned to an \\u003cstrong\\u003eunambiguous\\u003c/strong\\u003e primer version with zero mismatches. \\u0026ldquo;Mean Pair Coverage\\u0026rdquo; shows the literal primer-to-template (reference) alignment coverage across all best unambiguous primer-to-reference matches in a given taxon. \\u0026ldquo;3\\u0026prime;-end MM Forward\\u0026rdquo; displays the percentage of reference\\u0026ndash;primer matches with at least one 3\\u0026prime;-end mismatch for the forward primer, and \\u0026ldquo;3\\u0026prime;-end MM Reverse\\u0026rdquo; shows the same for the reverse primer.\\u003c/p\\u003e\\n\\u003ctable border=\\\"0\\\" cellspacing=\\\"0\\\" cellpadding=\\\"0\\\" width=\\\"712\\\"\\u003e\\n \\u003ctbody\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003eSpecies\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003ePerfect Match (%)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003eMean Pair Coverage (%)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e3\\u0026rsquo;-end MM Forward (%)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e3\\u0026rsquo;-end MM Reverse (%)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eAkkermansia muciniphila\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eBacillus subtilis\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e94.2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e99.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e2.4\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e1.9\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eBacteroides fragilis\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eBifidobacterium adolescentis\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eCandida albicans\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e92.9\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e99.6\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e2.9\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eClostridioides difficile\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e99.6\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.4\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eEscherichia coli\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e99.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eFaecalibacterium prausnitzii\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eFusobacterium nucleatum\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e97.8\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e99.9\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eLactobacillus fermentum\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e95.4\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e99.8\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e3.4\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.6\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eListeria monocytogenes\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e99.3\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.4\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eMethanobrevibacter smithii\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003ePrevotella corporis\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003ePseudomonas aeruginosa\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e95.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e99.8\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e1.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e1.5\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eRoseburia hominis\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eSaccharomyces cerevisiae\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e91.4\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e99.3\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e2.9\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eSalmonella enterica\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e99.7\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 147px;\\\"\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eVeillonella rogosae\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 102px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 173px;\\\"\\u003e\\n \\u003cp\\u003e100.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 157px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd valign=\\\"top\\\" style=\\\"width: 132px;\\\"\\u003e\\n \\u003cp\\u003e0.0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/tbody\\u003e\\n\\u003c/table\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eQuantitative Performance Metrics\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eDetection sensitivity and classification accuracy were assessed using the logarithmic and gut mock communities separately (Figure 7). In the logarithmic mock community, taxa above \\u003cem\\u003eca.\\u0026nbsp;\\u003c/em\\u003e1% relative abundance were reliably detected across all replicates, while those below \\u003cem\\u003eca.\\u0026nbsp;\\u003c/em\\u003e1% exhibited variable detection rates (Figure 7A). This indicates a steep detection drop off at low abundance levels, characteristic of primer and amplification bias in highly skewed communities. The gut mock community showed a similar trend, with taxa above \\u003cem\\u003eca.\\u0026nbsp;\\u003c/em\\u003e1% relative abundance being reliably detected, but there were two instances of a single repeat failing to detect an organism (Figure 7B).\\u003c/p\\u003e\\n\\u003cp\\u003ePrecision\\u0026ndash;recall analysis revealed strong classification performance across replicates. In the gut mock community, all replicates achieved high precision (\\u0026gt;0.8) and recall (\\u0026gt;0.8), with Q30+ samples (M2, M3) clustering near the identity line, indicating accurate and comprehensive taxon detection (Figure 7B). The lower quality replicate (M1, Q23) displayed a modest reduction in recall but maintained high precision, suggesting resilience of the classification strategy to reduced read quality. In contrast, the logarithmic mock community samples showed perfect precision (1.00) across all replicates, but substantially lower recall (0.50\\u0026ndash;0.70), reflecting the challenges of detecting very low abundance taxa rather than false positives (Figure 7A).\\u003c/p\\u003e\\n\\u003cp\\u003eTaken together, these results confirm that the pipeline delivers strong classification accuracy and consistent detection sensitivity across replicate samples. Genus level precision remained high even in lower quality data, with minimal false positives observed. Recall varied depending on taxon abundance and sample quality, with high abundance taxa (\\u0026gt;1% relative abundance) reliably detected across all replicates, and a consistent drop off below this level, particularly in the more skewed logarithmic mock. The lowest confidently detected taxon in the gut mock community was present at 2.8% relative abundance and recovered across all replicates, while in the logarithmic mock community, a taxon detected at 0.089% was recovered in only one of three replicates. This suggests that the empirical limit of detection for this pipeline lies between \\u003cstrong\\u003e0.1% and 1% relative abundance, but this cannot be accurately defined from the available data; at \\u0026ge;1% we observed high precision and recall across replicates. Together with precision and recall consistently \\u0026gt;0.85\\u003c/strong\\u003e, performance is in line with established performance expectations for amplicon-based microbiome profiling (Poncheewin\\u003cem\\u003e\\u0026nbsp;et al\\u003c/em\\u003e., 2020). Moreover, the observed precision and recall values compare favourably with Illumina-based pipelines such as NG-Tax and QIIME2 (Poncheewin \\u003cem\\u003eet al\\u003c/em\\u003e., 2020), highlighting the pipeline\\u0026rsquo;s reliability. These findings support the utility of this tool for accurate, reproducible, and domain-inclusive microbial community analysis across a range of sequencing conditions.\\u003c/p\\u003e\"},{\"header\":\"Discussion\",\"content\":\"\\u003cp\\u003eDespite increasing interest in Nanopore sequencing-based microbiome profiling, streamlined and flexible pipelines for long read, cross-domain amplicon analysis are in short supply. Existing tools such as EPI2ME lack the taxonomic and primer adaptability needed for modern studies. NAP was developed to address these gaps, enabling high throughput, domain-inclusive rRNA amplicon profiling across bacteria, archaea, and eukaryotes. NAP emphasises genus level accuracy while supporting species level resolution through careful interpretation. Key features, such as automated 16S/18S correction, primer-specific database pruning, and user-friendly configuration, make NAP both adaptable to diverse primers and accessible to researchers with minimal coding expertise.\\u003c/p\\u003e \\u003cp\\u003eTaxonomic dissimilarity metrics support the pipeline\\u0026rsquo;s reliability. Genus level Bray\\u0026ndash;Curtis dissimilarities in the logarithmic mock community were consistently low (\\u0026lt;\\u0026thinsp;0.13), within the normal variation reported for technical replicates (Bray\\u0026ndash;Curtis: ~0.12\\u0026thinsp;\\u0026plusmn;\\u0026thinsp;0.04; Yeh et al., \\u003cspan citationid=\\\"CR26\\\" class=\\\"CitationRef\\\"\\u003e2017\\u003c/span\\u003e). In the more complex gut mock community, Bray\\u0026ndash;Curtis values remained below 0.4, with higher variation attributable to sample-specific artefacts and known wet lab limitations. Jaccard dissimilarities, which are expected to be higher due to presence/absence sensitivity, typically remained\\u0026thinsp;\\u0026lt;\\u0026thinsp;0.33 across the mock communities, well within the 0.2\\u0026ndash;0.5 range reported as expected variance between other state-of-the-art pipelines (as a proxy for the variance expected within a single pipeline\\u0026rsquo;s outputs; O\\u0026rsquo;Sullivan et al., \\u003cspan citationid=\\\"CR15\\\" class=\\\"CitationRef\\\"\\u003e2021\\u003c/span\\u003e).\\u003c/p\\u003e \\u003cp\\u003eNAP demonstrated robust performance across sequencing conditions. Its adaptive filtering allowed accurate classification even in lower quality samples (e.g., Q20, \\u0026lt;\\u0026thinsp;100,000 reads), while high quality samples (Q30+, \\u0026gt;\\u0026thinsp;100,000 reads) produced highly consistent outputs. Reproducibility across technical replicates was high, with correlation coefficients (r ranging 0.68 to 1.0), concordance (\\u003cem\\u003ep\\u003c/em\\u003e\\u003csub\\u003e\\u003cem\\u003eCCC\\u003c/em\\u003e\\u003c/sub\\u003e ranging 0.53 to 1.0), and precision (1.0 in logarithmic, \\u003cem\\u003eca.\\u003c/em\\u003e 0.85 in gut mocks) aligning with benchmarks from Illumina-based tools (Poncheewin et al., \\u003cspan citationid=\\\"CR18\\\" class=\\\"CitationRef\\\"\\u003e2020\\u003c/span\\u003e) and, in some cases, even approaching short read metagenomics performance (Adams et al., \\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e). Recall in logarithmic mock communities was lower (\\u003cem\\u003eca.\\u003c/em\\u003e 0.5\\u0026ndash;0.7), but this is primarily attributable to primer bias and lysis inefficiencies, recognised limitations in amplicon workflows, rather than classifier error.\\u003c/p\\u003e \\u003cp\\u003eRelative abundance discrepancies were primarily driven by wet lab constraints. Underrepresentation of certain taxa (e.g., \\u003cem\\u003eAkkermansia\\u003c/em\\u003e and \\u003cem\\u003eBifidobacterium\\u003c/em\\u003e) aligned with known primer mismatches and lysis challenges. Species level artefacts were rare and traceable to either contaminating taxa or closely related sequences (e.g., \\u003cem\\u003eEscherichia\\u0026ndash;Shigella\\u003c/em\\u003e). The pipeline\\u0026rsquo;s conservative decontamination strategy eliminated most known contaminants, and corrected true positive abundances appropriately when inflated by contamination.\\u003c/p\\u003e \\u003cp\\u003eIn conclusion, NAP provides an accurate, reproducible, and open source solution for long read microbiome profiling. Its performance at the genus level is comparable to Illumina pipelines, under tested conditions, and species level accuracy is achievable with appropriate interpretation. Combined with its modular structure, primer-aware database integration, and low technical barrier, NAP fills a critical need in Nanopore amplicon sequencing workflows, supporting robust microbiome analysis across bacterial, archaeal, and eukaryotic domains, with modular components allowing primer-specific customisation.\\u003c/p\\u003e\"},{\"header\":\"Abbreviations\",\"content\":\"\\u003cdiv class=\\\"DefinitionList\\\"\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eASV\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003eamplicon sequence variant\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003ebp\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003ebase pair\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eBLAST\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003ebasic local alignment search tool\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eCD-HIT\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003ecluster database at high identity with tolerance\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eCV\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003ecoefficient of variation\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eDNA\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003edeoxyribonucleic acid\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003ePCoA\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003eprincipal coordinate analysis\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003ePCR\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003epolymerase chain reaction\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eQ-score (Phred score)\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003equality score representing probability of error in base calling\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003erRNA\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003eribosomal ribonucleic acid\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eSSU\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003esmall subunit (of rRNA)\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eTSS\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003etotal sum scaling\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eTSV\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003etab-separated values\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003cdiv class=\\\"DefinitionListEntry\\\"\\u003e \\u003cdiv class=\\\"Term\\\"\\u003eρc (ρCCC)\\u003c/div\\u003e \\u003cdiv class=\\\"Description\\\"\\u003e \\u003cp\\u003eLin\\u0026rsquo;s concordance correlation coefficient, β-diversity:beta diversity\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e \\u003c/div\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003ch4\\u003eAvailability of data and materials\\u003c/h4\\u003e\\n\\u003cp\\u003eThe datasets and code supporting the conclusions of this article are available as follows:\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eProject name: NAP\\u003c/p\\u003e\\n\\u003cp\\u003eProject home page: https://github.com/Luke-B-Jones/NAP\\u003c/p\\u003e\\n\\u003cp\\u003eArchived version: A permanently archived release is available \\u003cem\\u003evia\\u003c/em\\u003e Zenodo at DOI: 10.5281/zenodo.17662789\\u003c/p\\u003e\\n\\u003cp\\u003eOperating system(s): Platform-independent (tested on Linux)\\u003c/p\\u003e\\n\\u003cp\\u003eProgramming language: Python (≥3.9) and Bash\\u003c/p\\u003e\\n\\u003cp\\u003eOther requirements: Standard UNIX environment, and conda\\u003c/p\\u003e\\n\\u003cp\\u003eLicense: MIT license (open-source; free for academic and commercial use)\\u003c/p\\u003e\\n\\u003cp\\u003eRestrictions to use by non-academics: None.\\u003c/p\\u003e\\n\\u003cp\\u003eThe raw Nanopore amplicon sequencing data generated during this study are available in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA1367045.\\u003c/p\\u003e\\n\\u003ch4\\u003eCompeting interests\\u003c/h4\\u003e\\n\\u003cp\\u003eLuke Jones’s Ph.D. studentship is partly funded by Oxford Nanopore Technologies plc., which had no role in study design, data analysis, or manuscript preparation.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eFunding\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eWork funded by the University of Bath and Oxford Nanopore Technologies plc.\\u003c/p\\u003e\\n\\u003ch4\\u003eAuthors' contributions\\u003c/h4\\u003e\\n\\u003cp\\u003eLJ designed the study and implemented the pipeline, and carried out analyses. SB acquired funding and supervised the project. LJ and SB contributed to manuscript preparation. Both authors read and approved the final manuscript.\\u003c/p\\u003e\\n\\u003ch4\\u003e\\u003cem\\u003eAcknowledgements\\u003c/em\\u003e\\u003c/h4\\u003e\\n\\u003cp\\u003eWe thank Adrien Leger, Oxford Nanopore Technologies Ltd, for critical reading of the manuscript. We thank \\u003cstrong\\u003eJosephine Ilott\\u003c/strong\\u003e and \\u003cstrong\\u003eMorgan Cockrill\\u003c/strong\\u003e, Department of Life Sciences, University of Bath, for their contributions to the NAP pipeline: development of the decontamination module and improvement of the pipeline’s error reporting and overall robustness, respectively. Both contributions have been documented in the project repository.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\n \\u003cli\\u003eAdams, A.K. \\u003cem\\u003eet al.\\u003c/em\\u003e (2023) \\u0026ldquo;Qmatey: an automated pipeline for fast exact matching-based alignment and strain-level taxonomic binning and profiling of metagenomes,\\u0026rdquo; \\u003cem\\u003eBriefings in Bioinformatics\\u003c/em\\u003e, 24(6), p. bbad351. Available at: https://doi.org/10.1093/bib/bbad351.\\u003c/li\\u003e\\n \\u003cli\\u003eArvanitis, M. and Mylonakis, E. (2015) \\u0026ldquo;Fungal\\u0026ndash;bacterial interactions and their relevance in health,\\u0026rdquo; \\u003cem\\u003eCellular Microbiology\\u003c/em\\u003e, 17(10), pp. 1442\\u0026ndash;1446. Available at: https://doi.org/10.1111/cmi.12493.\\u003c/li\\u003e\\n \\u003cli\\u003eCamacho, C. \\u003cem\\u003eet al.\\u003c/em\\u003e (2009) \\u0026ldquo;BLAST+: architecture and applications,\\u0026rdquo; \\u003cem\\u003eBMC Bioinformatics\\u003c/em\\u003e, 10(1), p. 421. Available at: https://doi.org/10.1186/1471-2105-10-421.\\u003c/li\\u003e\\n \\u003cli\\u003eCharalampous, T. \\u003cem\\u003eet al.\\u003c/em\\u003e (2019) \\u0026ldquo;Nanopore metagenomics enables rapid clinical diagnosis of bacterial lower respiratory infection,\\u0026rdquo; \\u003cem\\u003eNature Biotechnology\\u003c/em\\u003e, 37(7), pp. 783\\u0026ndash;792. Available at: https://doi.org/10.1038/s41587-019-0156-5.\\u003c/li\\u003e\\n \\u003cli\\u003eDe Coster, W. \\u003cem\\u003eet al.\\u003c/em\\u003e (2018) \\u0026ldquo;NanoPack: visualizing and processing long-read sequencing data,\\u0026rdquo; \\u003cem\\u003eBioinformatics\\u003c/em\\u003e, 34(15), pp. 2666\\u0026ndash;2669. Available at: https://doi.org/10.1093/bioinformatics/bty149.\\u003c/li\\u003e\\n \\u003cli\\u003eFrey-Klett, P. \\u003cem\\u003eet al.\\u003c/em\\u003e (2011) \\u0026ldquo;Bacterial-Fungal Interactions: Hyphens between Agricultural, Clinical, Environmental, and Food Microbiologists,\\u0026rdquo; \\u003cem\\u003eMicrobiology and Molecular Biology Reviews\\u003c/em\\u003e, 75(4), pp. 583\\u0026ndash;609. Available at: https://doi.org/10.1128/mmbr.00020-11.\\u003c/li\\u003e\\n \\u003cli\\u003eGlassing, A. \\u003cem\\u003eet al.\\u003c/em\\u003e (2016) \\u0026ldquo;Inherent bacterial DNA contamination of extraction and sequencing reagents may affect interpretation of microbiota in low bacterial biomass samples,\\u0026rdquo; \\u003cem\\u003eGut Pathogens\\u003c/em\\u003e, 8(1), p. 24. Available at: https://doi.org/10.1186/s13099-016-0103-7.\\u003c/li\\u003e\\n \\u003cli\\u003eJanda, J.M. and Abbott, S.L. (2007) \\u0026ldquo;16S rRNA Gene Sequencing for Bacterial Identification in the Diagnostic Laboratory: Pluses, Perils, and Pitfalls,\\u0026rdquo; \\u003cem\\u003eJournal of Clinical Microbiology\\u003c/em\\u003e, 45(9), pp. 2761\\u0026ndash;2764. Available at: https://doi.org/10.1128/JCM.01228-07.\\u003c/li\\u003e\\n \\u003cli\\u003eKlindworth, A. \\u003cem\\u003eet al.\\u003c/em\\u003e (2013) \\u0026ldquo;Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies,\\u0026rdquo; \\u003cem\\u003eNucleic Acids Research\\u003c/em\\u003e, 41(1), p. e1. Available at: https://doi.org/10.1093/nar/gks808.\\u003c/li\\u003e\\n \\u003cli\\u003eLao, H.-Y. \\u003cem\\u003eet al.\\u003c/em\\u003e (2021) \\u0026ldquo;The clinical utility of two high-throughput 16S rRNA gene sequencing workflows for taxonomic assignment of unidentifiable bacterial pathogens in MALDI-TOF MS.\\u0026rdquo; bioRxiv, p. 2021.08.16.456588. Available at: https://doi.org/10.1101/2021.08.16.456588.\\u003c/li\\u003e\\n \\u003cli\\u003eLi, E. \\u003cem\\u003eet al.\\u003c/em\\u003e (2018) \\u0026ldquo;Benefits of antifungal therapy in asthma patients with airway mycosis: A retrospective cohort analysis,\\u0026rdquo; \\u003cem\\u003eImmunity, Inflammation and Disease\\u003c/em\\u003e, 6(2), pp. 264\\u0026ndash;275. Available at: https://doi.org/10.1002/iid3.215.\\u003c/li\\u003e\\n \\u003cli\\u003eLi, W. and Godzik, A. (2006) \\u0026ldquo;Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences,\\u0026rdquo; \\u003cem\\u003eBioinformatics\\u003c/em\\u003e, 22(13), pp. 1658\\u0026ndash;1659. Available at: https://doi.org/10.1093/bioinformatics/btl158.\\u003c/li\\u003e\\n \\u003cli\\u003eMcNichol, J. \\u003cem\\u003eet al.\\u003c/em\\u003e (2021) \\u0026ldquo;Evaluating and Improving Small Subunit rRNA PCR Primer Coverage for Bacteria, Archaea, and Eukaryotes Using Metagenomes from Global Ocean Surveys,\\u0026rdquo; \\u003cem\\u003emSystems\\u003c/em\\u003e, 6(3), pp. e00565-21. Available at: https://doi.org/10.1128/mSystems.00565-21.\\u003c/li\\u003e\\n \\u003cli\\u003e\\u0026ldquo;nanoporetech/dorado\\u0026rdquo; (2025). Oxford Nanopore Technologies. Available at: https://github.com/nanoporetech/dorado (Accessed: April 24, 2025).\\u003c/li\\u003e\\n \\u003cli\\u003eO\\u0026rsquo;Sullivan, D.M. \\u003cem\\u003eet al.\\u003c/em\\u003e (2021) \\u0026ldquo;An inter-laboratory study to investigate the impact of the bioinformatics component on microbiome analysis using mock communities,\\u0026rdquo; \\u003cem\\u003eScientific Reports\\u003c/em\\u003e, 11(1), p. 10590. Available at: https://doi.org/10.1038/s41598-021-89881-2.\\u003c/li\\u003e\\n \\u003cli\\u003ePalkova, L. et al. (2021) \\u0026ldquo;Evaluation of 16S rRNA primer sets for characterisation of microbiota in paediatric patients with autism spectrum disorder,\\u0026rdquo; Scientific Reports, 11, p. 6781. Available at: https://doi.org/10.1038/s41598-021-86378-w.\\u003c/li\\u003e\\n \\u003cli\\u003eParada, A.E., Needham, D.M. and Fuhrman, J.A. (2016) \\u0026ldquo;Every base matters: assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples,\\u0026rdquo; \\u003cem\\u003eEnvironmental Microbiology\\u003c/em\\u003e, 18(5), pp. 1403\\u0026ndash;1414. Available at: https://doi.org/10.1111/1462-2920.13023.\\u003c/li\\u003e\\n \\u003cli\\u003ePoncheewin, W. \\u003cem\\u003eet al.\\u003c/em\\u003e (2020) \\u0026ldquo;NG-Tax 2.0: A Semantic Framework for High-Throughput Amplicon Analysis,\\u0026rdquo; \\u003cem\\u003eFrontiers in Genetics\\u003c/em\\u003e, 10. Available at: https://doi.org/10.3389/fgene.2019.01366.\\u003c/li\\u003e\\n \\u003cli\\u003eQin, Y. \\u003cem\\u003eet al.\\u003c/em\\u003e (2023) \\u0026ldquo;Effects of error, chimera, bias, and GC content on the accuracy of amplicon sequencing,\\u0026rdquo; \\u003cem\\u003emSystems\\u003c/em\\u003e, 8(6), pp. e01025-23. Available at: https://doi.org/10.1128/msystems.01025-23.\\u003c/li\\u003e\\n \\u003cli\\u003eRamiro-Garcia, J. \\u003cem\\u003eet al.\\u003c/em\\u003e (2018) \\u0026ldquo;NG-Tax, a highly accurate and validated pipeline for analysis of 16S rRNA amplicons from complex biomes.\\u0026rdquo; F1000Research. Available at: https://doi.org/10.12688/f1000research.9227.2.\\u003c/li\\u003e\\n \\u003cli\\u003eRognes, T. \\u003cem\\u003eet al.\\u003c/em\\u003e (2016) \\u0026ldquo;VSEARCH: a versatile open source tool for metagenomics,\\u0026rdquo; \\u003cem\\u003ePeerJ\\u003c/em\\u003e, 4, p. e2584. Available at: https://doi.org/10.7717/peerj.2584.\\u003c/li\\u003e\\n \\u003cli\\u003eSalter, S.J. \\u003cem\\u003eet al.\\u003c/em\\u003e (2014) \\u0026ldquo;Reagent and laboratory contamination can critically impact sequence-based microbiome analyses,\\u0026rdquo; \\u003cem\\u003eBMC Biology\\u003c/em\\u003e, 12(1), p. 87. Available at: https://doi.org/10.1186/s12915-014-0087-z.\\u003c/li\\u003e\\n \\u003cli\\u003eShen, W. \\u003cem\\u003eet al.\\u003c/em\\u003e (2016) \\u0026ldquo;SeqKit: A Cross-Platform and Ultrafast Toolkit for FASTA/Q File Manipulation,\\u0026rdquo; \\u003cem\\u003ePLOS ONE\\u003c/em\\u003e, 11(10), p. e0163962. Available at: https://doi.org/10.1371/journal.pone.0163962.\\u003c/li\\u003e\\n \\u003cli\\u003eSilverman, J.D. \\u003cem\\u003eet al.\\u003c/em\\u003e (2021) \\u0026ldquo;Measuring and mitigating PCR bias in microbiota datasets,\\u0026rdquo; \\u003cem\\u003ePLOS Computational Biology\\u003c/em\\u003e, 17(7), p. e1009113. Available at: https://doi.org/10.1371/journal.pcbi.1009113.\\u003c/li\\u003e\\n \\u003cli\\u003eWang, H. \\u003cem\\u003eet al.\\u003c/em\\u003e (2023) \\u0026ldquo;Intestinal fungi and systemic autoimmune diseases,\\u0026rdquo; \\u003cem\\u003eAutoimmunity Reviews\\u003c/em\\u003e, 22(2), p. 103234. Available at: https://doi.org/10.1016/j.autrev.2022.103234.\\u003c/li\\u003e\\n \\u003cli\\u003eYeh, Y.-C. \\u003cem\\u003eet al.\\u003c/em\\u003e (2017) \\u0026ldquo;Taxon disappearance from microbiome analysis indicates need for mock communities as a standard in every sequencing run.\\u0026rdquo; bioRxiv, p. 206219. Available at: https://doi.org/10.1101/206219.\\u003c/li\\u003e\\n \\u003cli\\u003eYeh, Y.-C. \\u003cem\\u003eet al.\\u003c/em\\u003e (2021) \\u0026ldquo;Comprehensive single-PCR 16S and 18S rRNA community analysis validated with mock communities, and estimation of sequencing bias against 18S,\\u0026rdquo; \\u003cem\\u003eEnvironmental Microbiology\\u003c/em\\u003e, 23(6), pp. 3240\\u0026ndash;3250. Available at: https://doi.org/10.1111/1462-2920.15553.\\u003c/li\\u003e\\n \\u003cli\\u003eYoungblut, N.D. \\u003cem\\u003eet al.\\u003c/em\\u003e (2021) \\u0026ldquo;Vertebrate host phylogeny influences gut archaeal diversity,\\u0026rdquo; \\u003cem\\u003eNature Microbiology\\u003c/em\\u003e, 6(11), pp. 1443\\u0026ndash;1454. Available at: https://doi.org/10.1038/s41564-021-00980-2.\\u003c/li\\u003e\\n \\u003cli\\u003eZhang, B. \\u003cem\\u003eet al.\\u003c/em\\u003e (2020) \\u0026ldquo;Impact of bead-beating intensity on microbiome recovery in mouse and human stool: Optimization of DNA extraction.\\u0026rdquo; bioRxiv, p. 2020.06.15.151753. Available at: https://doi.org/10.1101/2020.06.15.151753.\\u003c/li\\u003e\\n \\u003cli\\u003eZhao, J., Rodriguez, J. and Martens-Habbena, W. (2023) \\u0026ldquo;Fine-scale evaluation of two standard 16S rRNA gene amplicon primer pairs for analysis of total prokaryotes and archaeal nitrifiers in differently managed soils,\\u0026rdquo; \\u003cem\\u003eFrontiers in Microbiology\\u003c/em\\u003e, 14. Available at: https://doi.org/10.3389/fmicb.2023.1140487.\\u003c/li\\u003e\\n \\u003cli\\u003e\\u003cem\\u003eZymoBIOMICS DNA/RNA Miniprep Kit\\u003c/em\\u003e (2025) \\u003cem\\u003eZymo Research International\\u003c/em\\u003e. Available at: https://zymoresearch.eu/products/zymobiomics-dna-rna-miniprep-kit (Accessed: April 24, 2025).\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":false,\"hideJournal\":false,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":false,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"bmc-bioinformatics\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":false,\"externalIdentity\":\"binf\",\"sideBox\":\"Learn more about [BMC Bioinformatics](http://bmcbioinformatics.biomedcentral.com/)\",\"snPcode\":\"\",\"submissionUrl\":\"https://www.editorialmanager.com/binf\",\"title\":\"BMC Bioinformatics\",\"twitterHandle\":\"@BMC_Bioinformatics\",\"acdcEnabled\":true,\"dfaEnabled\":false,\"editorialSystem\":\"em\",\"reportingPortfolio\":\"BMC Series\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":true},\"keywords\":\"Amplicon sequencing, Microbiome, Nanopore sequencing, Ribosomal RNA, Taxonomic classification\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-8173315/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-8173315/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003ch2\\u003eBackground\\u003c/h2\\u003e \\u003cp\\u003eNanopore sequencing offers a cost-effective and portable platform for amplicon-based microbiome analysis, but is still hindered by limited toolsets and sequencing error profile. While short-read technologies dominate microbial profiling workflows, their portability and flexibility are constrained. There is a need for robust pipelines tailored to Nanopore data that can support cross-kingdom ribosomal RNA profiling.\\u003c/p\\u003e\\u003ch2\\u003eResults\\u003c/h2\\u003e \\u003cp\\u003eWe introduce the Nanopore sequencing-based Amplicon Pipeline (NAP; \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://github.com/Luke-B-Jones/NAP\\u003c/span\\u003e\\u003cspan address=\\\"https://github.com/Luke-B-Jones/NAP\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e), an open-source tool optimised for flexible, mixed-domain primer sets (such as 515Y/926R). NAP performs quality filtering, chimera removal, centroid identification, and BLAST-based taxonomic classification with consensus correction. It outputs normalised, bias-corrected tab-separated value files suitable for downstream analysis. Validation against two commercial mock communities showed that NAP achieves genus-level precision of up to 100%, with taxonomic concordance comparable to Illumina-based workflows. Detection sensitivity was consistently reliable above 1% relative abundance. β-diversity measures, including Bray\\u0026ndash;Curtis and Jaccard indices, fell within expected replicate variation. Taxonomic agreement remained high across a range of read depths and sequencing qualities, with most errors attributable to laboratory-derived artefacts rather than computational limitations.\\u003c/p\\u003e\\u003ch2\\u003eConclusions\\u003c/h2\\u003e \\u003cp\\u003eNAP delivers robust genus-level performance on par with Illumina workflows, with the potential to achieve species-level resolution using longer amplicons. Its compatibility with portable and cost-effective sequencing makes it well suited for accurate long-read microbiome profiling in both laboratory and field environments.\\u003c/p\\u003e\",\"manuscriptTitle\":\"NAP: An Open-Source Pipeline for Cross-Domain Microbiome Profiling Using Nanopore Sequencing-Derived Amplicon Data\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2026-03-10 13:19:52\",\"doi\":\"10.21203/rs.3.rs-8173315/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0},{\"type\":\"decision\",\"content\":\"Revision requested\",\"date\":\"2026-04-08T10:37:19+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorInvitedReview\",\"content\":\"\",\"date\":\"2026-04-02T15:18:44+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"editorInvitedReview\",\"content\":\"\",\"date\":\"2026-03-25T08:35:07+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"79500813441640693214710135268365119309\",\"date\":\"2026-03-19T08:36:38+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"5624464416678305925457449970119980788\",\"date\":\"2026-03-16T06:54:19+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewersInvited\",\"content\":\"\",\"date\":\"2026-03-05T02:28:05+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorInvited\",\"content\":\"\",\"date\":\"2026-03-03T17:10:59+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorAssigned\",\"content\":\"\",\"date\":\"2025-11-26T08:48:18+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"checksComplete\",\"content\":\"\",\"date\":\"2025-11-26T08:43:08+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"submitted\",\"content\":\"BMC Bioinformatics\",\"date\":\"2025-11-21T11:45:25+00:00\",\"index\":\"\",\"fulltext\":\"\"}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"bmc-bioinformatics\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":false,\"externalIdentity\":\"binf\",\"sideBox\":\"Learn more about [BMC Bioinformatics](http://bmcbioinformatics.biomedcentral.com/)\",\"snPcode\":\"\",\"submissionUrl\":\"https://www.editorialmanager.com/binf\",\"title\":\"BMC Bioinformatics\",\"twitterHandle\":\"@BMC_Bioinformatics\",\"acdcEnabled\":true,\"dfaEnabled\":false,\"editorialSystem\":\"em\",\"reportingPortfolio\":\"BMC Series\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":true}}],\"origin\":\"\",\"ownerIdentity\":\"f2266557-5d2c-410f-a594-f9c475064ed6\",\"owner\":[],\"postedDate\":\"March 10th, 2026\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"under-review\",\"subjectAreas\":[],\"tags\":[],\"updatedAt\":\"2026-05-04T09:24:29+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2026-03-10 13:19:52\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-8173315\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-8173315\",\"identity\":\"rs-8173315\",\"version\":[\"v1\"]},\"buildId\":\"XKTyCvWXoU3ODBz1xrDgd\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}