BMC Bioinformatics最新文献

{"title":"RIFTT: Robust Identification of Fusion genes in Tumor Transcriptomes.","authors":"Paul Kerbs, Lisa Latzko, Ulrich Mansmann, Kristian Unger, Philipp A Greif, Sebastian Vosberg","doi":"10.1186/s12859-026-06725-4","DOIUrl":"10.1186/s12859-026-06725-4","url":null,"abstract":"<p><strong>Background: </strong>Fusion genes serve as crucial biomarkers for cancer diagnosis, prognosis, and therapy, yet many detection tools lead to high false-positive rates and poor consensus. We developed RIFTT, a pipeline for Robust Identification of Fusion genes in Tumor Transcriptomes which integrates Arriba and FusionCatcher, two state-of-the-art fusion callers, together with novel scoring metrics (Promiscuity Score, Fusion Transcript Score, and Robustness Score).</p><p><strong>Results: </strong>Benchmarking on TCGA-LAML, BeatAML, and LL-100 datasets demonstrated RIFTT's ability to reduce the number of candidates by up to 95% compared to single callers. At the same time, it preserves high-confidence events, outperforming both the nf-core/rnafusion pipeline and an Arriba/FusionCatcher consensus approach. RIFTT also identified novel high-evidence candidates, including a previously unreported recurrent fusion in acute myeloid leukemia.</p><p><strong>Conclusion: </strong>Packaged in a Singularity container for ease of use across computing environments, RIFTT provides standardized visual and annotated tabular outputs to support downstream analysis in research workflows and clinical molecular diagnostics. RIFTT is freely available at https://github.com/pkerbs/RIFTT .</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13729477/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149317647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PoolParty: streamlined design of DNA sequence libraries in Python.","authors":"Zhihan Liu, Aidan Cordero, Justin B Kinney","doi":"10.1186/s12859-026-06696-6","DOIUrl":"10.1186/s12859-026-06696-6","url":null,"abstract":"<p><strong>Background: </strong>Computationally designed DNA sequence libraries are essential components of massively parallel reporter assays (MPRAs), deep mutational scanning (DMS) experiments, and other multiplex assays of variant effect (MAVEs). They are also increasingly used in silico to analyze genomic AI models. Designing these libraries, however, remains tedious and error-prone due to the scarcity of purpose-built software.</p><p><strong>Results: </strong>Here we describe PoolParty, a Python package that streamlines the design of complex oligo pools using a simple but flexible API. In PoolParty, each library is represented by a computational graph that can be specified in just a few lines of code. Over 50 built-in operations cover nucleotide- and codon-level mutagenesis, motif insertion, barcode generation, and more. PoolParty automatically generates informative names for each sequence and provides \"design cards\" detailing how each sequence was generated. Visualization methods let users quickly audit library content and inspect the underlying graph. PoolParty thus transforms oligo pool design from a tedious task requiring custom functions and scripts into a structured, transparent, and reproducible process.</p><p><strong>Conclusions: </strong>PoolParty streamlines the design of DMS, MPRA, and other multiplex assay libraries, and the design cards it provides can help researchers systematically probe and interpret genomic AI models. PoolParty can also be extended to support new assays and analysis strategies as they emerge.</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13723123/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149294337","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Attentive interpretable tabular learning (TabNet) for miRNA expression classification in cervical cancer and Alzheimer's disease.","authors":"Hilal Haznedar, Gözde Özsert Yiğit, Bülent Haznedar","doi":"10.1186/s12859-026-06706-7","DOIUrl":"10.1186/s12859-026-06706-7","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background and objective: &lt;/strong&gt;The classification of RNA sequencing (RNA-seq) data presents a fundamental challenge in computational biomedicine: raw read counts are high-dimensional, over-dispersed, and heteroscedastic, requiring careful preprocessing before machine learning models can be applied. Existing deep learning approaches for RNA-seq classification have largely overlooked the tabular nature of genomic expression data and have offered limited embedded interpretability. This study proposes TabNet, an attentive interpretable tabular learning architecture, as a framework for miRNA-based disease classification and evaluates its performance and interpretability against three established baseline models.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;Two publicly available miRNA expression datasets were examined: a cervical cancer cohort (58 samples, 714 miRNAs) and an Alzheimer's disease cohort (70 samples, 503 miRNAs). Raw count data were preprocessed following the MLSeq framework, encompassing DESeq median-ratio normalisation, near-zero variance filtering, supervised gene ranking based on DESeq differential expression analysis, and the DESeq variance-stabilising transformation (vst). Gene subsets corresponding to 5%, 10%, and 30% of the post-filtering feature space were evaluated. TabNet was benchmarked against eight baselines spanning four learning paradigms: a 1D Convolutional Neural Network (CNN), a Long Short-Term Memory network (LSTM), Extreme Gradient Boosting (XGBoost), Random Forest, ElasticNet-regularised logistic regression, a Multi-Layer Perceptron (MLP), the prior-fitted transformer TabPFN, and the inherently interpretable Explainable Boosting Machine (EBM). Each configuration was assessed over 50 independent random train/test splits (70%/30%) under a strict leakage-free protocol in which all preprocessing, hyperparameter selection, and early stopping relied exclusively on inner validation subsets of the training partitions; all metrics are reported as mean &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; SD.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Under the leakage-free protocol, a group of four methods reached a statistically indistinguishable accuracy ceiling: on the cervical cancer dataset, ElasticNet (95.06% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 4.66), EBM (94.94% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 4.60), TabPFN (94.71% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 4.64), and Random Forest (94.35% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 5.82); on the Alzheimer's disease dataset, TabPFN (70.86% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 10.26), EBM (70.67% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 12.67), and ElasticNet (70.48% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 9.76) led. TabNet reached 85.18% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 11.10 (cervical) and 64.38% &lt;math&gt;&lt;mo&gt;±&lt;/mo&gt;&lt;/math&gt; 11.23 (Alzheimer's) at its respective best levels -- below the top tier, but with the distinctive property of producing sparse, instance-level attention explanations as part of its forward computation. Supplementary analyses showed that self-supervised pre-training conferred no significant benefit at ","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13737494/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149340611","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Extracting candidate motif patterns for multiple RNA modifications using direct RNA sequencing.","authors":"Nicolò Alagna, Johannes Miedema, Yahia Basiony, Susanne Gerber","doi":"10.1186/s12859-026-06724-5","DOIUrl":"10.1186/s12859-026-06724-5","url":null,"abstract":"<p><p>RNA modifications have emerged as key regulators of virtually all aspects of RNA biology, driving growing interest in computational approaches for their transcriptome-wide detection. For some RNA modifications, the search space can be reduced by exploiting characteristic sequence motifs, as exemplified by the canonical DRACH motif associated with m⁶A. However, comparable sequence preferences remain largely unknown for most other RNA modifications, making systematic motif discovery considerably more challenging. Here, we applied exponential population analysis to direct RNA sequencing data from HEK293T cells and peripheral human blood to identify candidate motif patterns for pseudouridine (Ψ), inosine (I), and 5-methylcytosine (m⁵C). Our results demonstrate that exponential population analysis provides a systematic framework for identifying candidate motif patterns from nanopore-derived RNA modification predictions, thereby prioritizing sequence contexts for future biological validation.</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13700448/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149262164","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ENflorA: a flexible tool for the bulk submission of next generation sequencing data to the European Nucleotide Archive.","authors":"Antonio Alfaro de Prado Martín, Francisco Fajardo-Gutiérrez, Astrid de Mestier, Katja Reichel","doi":"10.1186/s12859-026-06714-7","DOIUrl":"10.1186/s12859-026-06714-7","url":null,"abstract":"<p><strong>Background: </strong>With the advancement of sequencing technologies and increasing quality of assembly and annotation tools, the amount of biological sequence data produced by research grows rapidly. To keep this data accessible for potential re-analysis, it is typically stored in public online databases, such as the European Nucleotide Archive (ENA). However, the increase in data entails an increase in time needed for metadata curation and upload. We propose a tool to simplify and automatize the upload of experimental and metadata to public repositories.</p><p><strong>Results: </strong>We developed ENflorA, a set of scripts by which sample, sequence read and (annotated) assembly data can be uploaded to ENA in a simple and standardized manner. Our software allows easy metadata entry via spreadsheets, is suitable for batch upload of data from different sources, and can be fully integrated in bioinformatic workflows. It is platform independent and can be run natively on High Performance Computers, integrates an lftp upload option for big files and a test upload option.</p><p><strong>Conclusions: </strong>By easing the ENA submission process, ENflorA contributes to making high-throughput sequencing data findable, accessible, interoperable and reusable.</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13689177/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149236737","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Wild-AC: A fast, index-free multi-pattern string matching algorithm with wildcard support for proteomics.","authors":"Chris Bielow","doi":"10.1186/s12859-026-06686-8","DOIUrl":"10.1186/s12859-026-06686-8","url":null,"abstract":"<p><strong>Background: </strong>Efficient multi-pattern string matching with wildcard support is essential in proteomics, where peptides must be mapped to protein databases containing ambiguous amino acids (e.g., B, Z, X). Existing approaches either require preprocessing (e.g., FM-index) or fail to handle wildcards in the text.</p><p><strong>Results: </strong>We present Wild-AC, a modified Aho-Corasick algorithm that supports wildcards in the text by branching the search into parallel 'scout' paths, one per possible wildcard representation, while the unmodified primary search continues unaffected in the common, wildcard-free case. Wild-AC outperforms the FM-index (wildcard case) and matches or exceeds Wu-Manber (exact search) in speed for realistic proteomics workloads, i.e., 1000-500 000 peptide patterns (average length ∼18 amino acids) searched against protein databases (texts) of 3 × 10⁶-2.1 × 10⁸ characters. On databases masked to a wildcard rate of 5%, Wild-AC retains its advantage for large peptide sets, while the FM-index becomes preferable for small ones. It requires no index, scales well with pattern count, and supports multi-threading. The C++ implementation is open source at https://github.com/Wild-AC/Wild-AC .</p><p><strong>Conclusion: </strong>Wild-AC extends the Aho-Corasick algorithm to support wildcards and outperforms FM-Index and Wu-Manber in the task of peptide-protein mapping while considering ambiguous amino acids.</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13659980/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149164265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multivariate GWAS framework for family trios with parental phenotypes to control dynastic effects.","authors":"Shun Zhang, Jia-Hao Mai, Qi Zhong, Qiu-Wen Zhu, Yuan-Sheng Li, Xian-Bo Wu, Ji-Yuan Zhou","doi":"10.1186/s12859-026-06597-8","DOIUrl":"10.1186/s12859-026-06597-8","url":null,"abstract":"<p><strong>Background: </strong>Effect size estimates in genome-wide association studies (GWAS) and Mendelian randomization (MR) based on unrelated individuals are often confounded by dynastic effects (DE). Existing family-trio-based methods can mitigate this bias, but they typically discard parental phenotypes and cannot jointly model multiple correlated traits. We propose FT-SEM, a multivariate GWAS framework that uses structural equation modeling for family trio data, with the aim of integrating genotypes and phenotypes from parents and offspring across multiple traits to estimate direct genetic effects on a shared latent factor.</p><p><strong>Results: </strong>Simulations demonstrated that FT-SEM yields unbiased effect estimates with well-controlled type I error, while improving statistical power over conventional trio-based models by incorporating parental phenotypic information. In a genome-wide scan of 778 UK Biobank family trios, no locus reached genome-wide significance, but the FT-SEM-derived family-corrected summary statistics for a latent obesity factor provided an unbiased outcome base for downstream MR. Two-sample MR analyses revealed that the apparent association between systolic blood pressure and obesity was substantially attenuated after accounting for familial confounding, suggesting that it may be spurious. In contrast, the effect of high-density lipoprotein cholesterol on obesity remained directionally consistent and robust, supporting a genuine causal relationship.</p><p><strong>Conclusions: </strong>FT-SEM provides a rigorous and efficient multivariate framework that eliminates dynastic bias while maximizing the utility of family-trio data. The resulting summary statistics can be seamlessly integrated into two-sample MR analyses, highlighting the necessity of family-trio-informed multivariate modeling for guarding against spurious causal claims in complex trait genetics.</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-09-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13617700/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149073173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"meth-SemiCancer2: a cancer subtyping framework leveraging domain adaptation with contrastive learning for batch effect correction in DNA methylation profiles.","authors":"Joung Min Choi, Heejoon Chae","doi":"10.1186/s12859-026-06656-0","DOIUrl":"10.1186/s12859-026-06656-0","url":null,"abstract":"<p><strong>Background: </strong>Cancer subtyping is essential for precision oncology, enabling early diagnosis, targeted therapy, and improved prognosis. DNA methylation is a powerful molecular marker for subtype classification, but its application is hindered by high dimensionality, batch effects, and the scarcity of subtype-labeled datasets. We previously developed meth-SemiCancer, a semi-supervised framework that leverages unlabeled methylation data to improve subtype prediction. While effective, meth-SemiCancer did not explicitly address batch effect correction or robust representation learning, limiting its generalizability across cohorts. Without addressing these challenges, technical noise can obscure true biological signals, leading to spurious subtype boundaries and limiting reproducibility across studies.</p><p><strong>Results: </strong>To overcome these limitations, we developed meth-SemiCancer2, an enhanced framework that employs domain adaptation coupled with contrastive learning, followed by semi-supervised fine-tuning with subtype alignment. Specifically, meth-SemiCancer2 first pretrains the model on labeled source datasets to capture subtype-specific patterns, then applies adversarial training to align distributions between source and unlabeled target cohorts. Contrastive learning is subsequently introduced to strengthen domain-invariant feature extraction, and finally, semi-supervised fine-tuning with subtype alignment iteratively refines pseudo-labels to improve subtype separation across cohorts. Through this multi-phase, the model corrects batch effects while learning robust and discriminative subtype representations from high-dimensional methylation profiles.</p><p><strong>Conclusion: </strong>Evaluations across seven cancer types-breast cancer, colon adenocarcinoma, prostate adenocarcinoma, glioblastoma, renal cell carcinoma, thyroid carcinoma, and low-grade glioma-demonstrated that meth-SemiCancer2 consistently outperformed recently proposed deep learning-based DNA methylation subtype classification frameworks, including our previous meth-SemiCancer model, a representative domain adaptation method, two state-of-the-art DNA methylation batch correction approaches, and conventional machine learning classifiers. In labeled source datasets, 10-fold cross-validation verified its superior accuracy and reliability in subtype prediction. When applied to target cohorts, the model generalized effectively, producing accurate subtype assignments with clearly delineated clusters. Our evaluations further indicated that meth-SemiCancer2 sustained stable performance even when only fractions of the source or target data were available, highlighting its utility under data-limited conditions. meth-SemiCancer2 is publicly accessible at https://github.com/cbi-bioinfo/meth-semicancer2 .</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-09-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13693024/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149242796","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CADD-ThresholdApp: an interactive web application for calibrating CADD scores using ClinVar.","authors":"Cora Leifheit, Martin Kircher, Max Schubach","doi":"10.1186/s12859-026-06675-x","DOIUrl":"10.1186/s12859-026-06675-x","url":null,"abstract":"<p><strong>Background: </strong>The Combined Annotation Dependent Depletion (CADD) score is one of the most widely used genome-wide variant deleteriousness scores. In practice, however, the distribution of CADD scores differs across genes and gene sets and may also vary between genome builds and CADD releases, complicating the choice of a single threshold for prioritizing potentially pathogenic variants in routine diagnostics.</p><p><strong>Results: </strong>We present CADD-ThresholdApp, an interactive web application that supports threshold selection by visualizing CADD-score distributions of clinically classified variants from ClinVar and by reporting threshold-dependent performance metrics. The app enables comparisons across genome releases (GRCh37/GRCh38) and CADD versions (v1.6/v1.7), supports user-defined genes and gene lists, as well as curated gene panels from PanelApp. CADD-ThresholdApp is implemented in a modular, extensible framework and can be adapted to additional variant resources and deleteriousness scores. The app is available as a hosted web application at https://cadd-threshold.kircherlab.bihealth.org/ and can be installed locally via container, PyPI, Bioconda, or from source.</p><p><strong>Conclusions: </strong>Overall, CADD-ThresholdApp provides a practical and transparent framework for context-specific threshold calibration to support more informed variant interpretation in research and diagnostics.</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13617862/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149073190","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"POMS enhances open spectral library search for identification of modified peptides.","authors":"Younghee Seo, Eunok Paek, Seungjin Na","doi":"10.1186/s12859-026-06672-0","DOIUrl":"10.1186/s12859-026-06672-0","url":null,"abstract":"<p><strong>Background: </strong>Peptide identification from tandem mass spectrometry (MS/MS) data is a central task in proteomics. Spectral library searching combined with open modification search (OMS) has emerged as an effective strategy for identifying peptides carrying unexpected post-translational modifications (PTMs). However, existing methods do not adequately account for modification-induced fragment-ion shifts during candidate selection, leading to missed identifications when mass shifts distort spectral similarity.</p><p><strong>Results: </strong>We present POMS, a modification-aware framework for open spectral library search that improves candidate retrieval by integrating complementary spectral representations with sequence-derived theoretical fragment features. These representations compensate for modification-induced fragment-ion shifts and improve alignment between query and library spectra, increasing robustness to modification site variability. Benchmarking on large-scale human MS/MS datasets showed that POMS yielded up to ~ 8% more peptide identifications during the open search phase than conventional methods. Cross-validation with independent database search engines further demonstrated a > 5% increase in consistent peptide identifications. Notably, POMS substantially improved the identification of peptides carrying C-terminal modifications, for which conventional candidate retrieval methods are particularly susceptible to fragment-ion shifts.</p><p><strong>Conclusions: </strong>POMS improves the sensitivity and reliability of modification-tolerant spectral library searching by incorporating modification-aware candidate retrieval while preserving computational efficiency. The framework provides a practical solution for large-scale PTM discovery and is readily applicable to existing spectral library search workflows. POMS is publicly available under the CC BY-NC-SA 4.0 license at https://github.com/icp-kbsi/POMS .</p>","PeriodicalId":8958,"journal":{"name":"BMC Bioinformatics","volume":"27 1","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13741594/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149340612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}