Briefings in bioinformatics最新文献

{"title":"AdaSemb: an adaptive knowledge-driven deep learning framework integrating cancer protein assemblies for predicting PI3Kα inhibitor response and resistance.","authors":"Zaiduo Li, Qiang Yang, Long Xu, Weihe Dong, Xiaochuan Yang, Xianyu Zhang, Tiansong Yang, Xiaokun Li, Mingliang Liu","doi":"10.1093/bib/bbaf510","DOIUrl":"10.1093/bib/bbaf510","url":null,"abstract":"<p><p>Protein kinases regulate diverse cellular functions, including cell cycle progression, metabolism, differentiation, and survival, with their dysregulation implicated in multiple carcinogenic processes. Phosphatidylinositol 3-kinase alpha inhibitors (PI3K$ alpha $is) have revolutionized breast cancer treatment, but acquired resistance remains a major clinical challenge, with around 40% of patients experiencing progression within 4-6 months. Current drug response prediction (DRP) methods typically rely on individual pathways or biomarkers, limiting their ability to capture complex cancer-specific molecular interactions and predict resistance mechanisms. To overcome these limitations, we present AdaSemb, an adaptive, knowledge-driven deep learning framework that uses a multi-protein assembly map to predict responses and resistance to PI3K$ alpha $i. AdaSemb comprises two modules: the AdaSemb-PA module incorporates tumor genomic variations into a biological structural neural network, while the AdaSemb-DRP module uses conditional domain adversarial networks to enhance gene-drug distribution generalization. By combining genomic data with drug molecular structures, AdaSemb identifies critical protein combinations linked to drug resistance. In validation with 1244 cancer cell lines and patient-derived xenografts (PDX), AdaSemb outperformed existing DRP models. In a cohort of 116 breast cancer patients from the Cancer Genome Atlas (TCGA), it predicted significantly longer survival for sensitive patients, surpassing traditional biomarkers in precision. Furthermore, we identified seven key assemblages that integrate mutations from 93 genes, which distinguish alpelisib sensitive and resistant cell lines. These results are applicable to breast cancer patient samples and PDX models, demonstrating AdaSemb's significant clinical potential in personalized treatment and prediction of resistance for breast cancer.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12477610/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145184653","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"scHSC: enhancing single-cell RNA-seq clustering via hard sample contrastive learning.","authors":"Sheng Fang, Xiaokang Yu, Xinyi Xu, Jingxiao Zhang, Xiangjie Li","doi":"10.1093/bib/bbaf485","DOIUrl":"10.1093/bib/bbaf485","url":null,"abstract":"<p><p>Single-cell RNA sequencing (scRNA-seq) provides high-throughput information about the genome-wide gene expression levels at the single-cell resolution, bringing a precise understanding on the transcriptome of individual cells. Unfortunately, the rapidly growing scRNA-seq data and the prevalence of dropout events pose substantial challenges for clustering and cell type annotation. Here, we propose a deep learning method, scHSC, that employs hard sample mining through contrastive learning for clustering scRNA-seq data. Focusing on hard samples, this approach simultaneously integrates gene expression and topological structure information between cells to improve clustering accuracy. By adjusting the weights of hard positive and hard negative samples during the iterative training process, scHSC employs an adaptive weighting strategy to integrate contrastive learning with a ZINB model for single-cell clustering tasks. Extensive experiments on 18 single-cell RNA-seq real datasets demonstrate that scHSC exhibits significant superiority in clustering performance compared to existing deep learning-based clustering methods. scHSC is implemented in Python based on the PyTorch framework. The source code and datasets are available via https://github.com/fangs25/scHSC.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12451106/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145112009","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-dimensional computational framework of drug-induced hepatotoxicity: integrating molecular structure features with disease pathogenesis.","authors":"Huayu Zhong, Juanji Wang, Xiaoxiao Liu, Xiaoyun Wei, Chengcheng Zhou, Taiyan Zou, Xin Han, Lingyun Mo, Wenling Qin, Yonghong Zhang","doi":"10.1093/bib/bbaf456","DOIUrl":"10.1093/bib/bbaf456","url":null,"abstract":"<p><p>Drug-induced hepatotoxicity (DIH), characterized by diverse phenotypes and complex mechanisms, remains a critical challenge in drug discovery. To systematically decode this diversity and complexity, we propose a multi-dimensional computational framework integrating molecular structure analysis with disease pathogenesis exploration, focusing on drug-induced intrahepatic cholestasis (DIIC) as a representative DIH subtype. First, a graph-based modularity maximization algorithm identified DIIC risk genes, forming a DIIC module and eight disease pathogenesis clusters. Network proximity values between drug targets and DIIC clusters were calculated to define drug-disease relationships. Subsequently, a random forest model combining Mordred molecular descriptors, structural alerts (SAs), and network proximity achieved robust DIIC prediction: Accuracy(ACC) = 0.740 ± 0.014 and area under the curve (AUC) = 0.828 ± 0.008 (ntraining = 342, nvalidation = 114, nexternal test = 295, randomly modeling 100 times). Notably, a K-nearest neighbors-graph convolutional network classified drugs into 8 clusters, with the Cluster 3 model demonstrating superior performance (ACC = 0.810 ± 0.024; AUC = 0.890 ± 0.014; ntraining = 186, nvalidation = 63, nexternal test = 172). Mechanistic analysis linked critical SAs to DIIC pathogenesis: (i) Furan (SA3) perturbed cytochrome P450-mediated metabolism and regulation of lipid metabolism by PPARα; (ii) Nitrogen-sulfur heteroatom chains (SA7) disrupted metabolism of steroids; (iii) Phenylthio groups (SA12) and their CYP450 metabolites induced cholestasis. This multi-dimensional framework bridges molecular features and disease mechanisms, offering a generalizable strategy for toxicity prediction and pathway-centric drug safety evaluation, especial for complex disease.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12415848/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145013795","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"OligoY pipeline for full Y chromosome painting.","authors":"Isabela Almeida, Henry A B Bruno, Eduardo Guimarães Dupim, Mara Maria L Santana Pinheiro, Antonio Bernardo Carvalho, Maria D Vibranovski","doi":"10.1093/bib/bbaf507","DOIUrl":"10.1093/bib/bbaf507","url":null,"abstract":"<p><p>The Y chromosome's unique structure poses challenges for cytogenetic studies, especially in designing probes for FISH Oligopaint labeling experiments (Fluorescence in situ hybridization). The standard protocol for designing probes for these experiments discards repetitive sequences to avoid off-target hybridization. Given the highly repetitive nature of the Y chromosome, assemblies often remain fragmented, leaving significant regions incompletely sequenced, unprobed, and poorly understood. Among these, the remaining nonrepetitive sequences are usually insufficient to design probes and efficiently perform FISH Oligopaint assays, since they do not cover most regions of the chromosome. This limitation hinders comprehensive cytogenetic studies, which are crucial not only for understanding the Y chromosome's role in genetics but also for broader applications in evolutionary biology, medicine, and conservation. Here, we introduce a new computational pipeline to design full chromosome fluorescent labeling probes for the Y chromosome of any species of interest. Based on open-source tools, the OligoY pipeline increases the amount of contigs assigned to the Y chromosome from the reference genome assembly, and effectively uses repetitive sequences unique to the target chromosome to design probes. Throughout its steps, the pipeline gives the user the autonomy to choose parameters, maximizing the overall efficiency of cytogenetic experiments. After extensive in silico and in situ testing and validation with the human and Drosophila melanogaster genomes, we show for the first time a pipeline for FISH Oligopaint probe design that significantly increases previous Y chromosome staining with no off-target signal.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12481694/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145198442","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"scSorterDL: a deep neural network-enhanced ensemble LDAs for single cell classifications.","authors":"Kailun Bai, Belaid Moa, Xiaojian Shao, Xuekui Zhang","doi":"10.1093/bib/bbaf446","DOIUrl":"10.1093/bib/bbaf446","url":null,"abstract":"<p><p>The emergence of single-cell RNA sequencing (scRNA-seq) technology has transformed our understanding of cellular diversity, yet it presents notable challenges for cell type annotation due to data's high dimensionality and sparsity. To tackle these issues, we present scSorterDL, an innovative approach that combines penalized Linear Discriminant Analysis (pLDA), swarm learning, and deep neural networks (DNNs) to improve cell type classification. In scSorterDL, we generate numerous random subsets of the data and apply pLDA models to each subset to capture varied data aspects. The model outputs are then consolidated using a DNN that identifies complex relationships among the pLDA scores, enhancing classification accuracy by considering interactions that simpler methods might overlook. Utilizing GPU computing for both swarm learning and deep learning, scSorterDL adeptly manages large datasets and high-dimensional gene expression data. We tested scSorterDL on 13 real scRNA-seq datasets from diverse species, tissues, and platforms, as well as on 20 pairs of cross-platform datasets. Our method surpassed nine current cell annotation tools in both accuracy and robustness, indicating exceptional performance in both cross-validation and cross-platform contexts. These findings underscore the potential of scSorterDL as an effective and adaptable tool for automated cell type annotation in scRNA-seq research. The code is available on GitHub: https://github.com/kellen8hao/scSorterDL.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12400813/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144943421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Phage quest: a beginner's guide to explore viral diversity in the prokaryotic world.","authors":"Carolin Charlotte Wendling, Marie Vasse, Sébastien Wielgoss","doi":"10.1093/bib/bbaf449","DOIUrl":"10.1093/bib/bbaf449","url":null,"abstract":"<p><p>The increasing interest in finding new viruses within (meta)genomic datasets has fueled the development of computational tools for virus detection and characterization from environmental samples. One key driver is phage therapy, the treatment of drug-resistant bacteria with tailored bacteriophage cocktails. Yet, keeping up with the growing number of automated virus detection and analysis tools has become increasingly difficult. Both phage biologists with limited bioinformatics expertise and bioinformaticians with little background in virus biology will benefit from this guide. It focuses on navigating routine tasks and tools related to (pro)phage detection, gene annotation, taxonomic classification, and other downstream analyses. We give a brief historical overview of how detection methods evolved, starting with early sequence-composition assessments to today's powerful machine-learning and deep learning techniques, including emerging language models capable of mining large, fragmented, and compositionally diverse metagenomic datasets. We also discuss tools specifically aimed at detecting filamentous phages (Inoviridae), a challenge for most phage predictors. Rather than providing an exhaustive list, we emphasize actively maintained and state-of-the-art tools that are accessible via web or command-line interfaces. This guide provides basic concepts and useful details about automated phage analysis for researchers in different biological and medical disciplines, helping them choose and apply appropriate tools for their quest to explore the genetic diversity and biology of the smallest and most abundant replicators on Earth.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12406692/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144943455","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Clinical and data-driven optimization of Genomiser for rare disease patients: experience from the Hong Kong Genome Project.","authors":"Anson Man Chun Xi, Denis Long Him Yeung, Wei Ma, Dingge Ying, Amy Hin Yan Tong, Dicky Or, Shirley Pik Ying Hue, Hong Kong Genome Project, Annie Tsz-Wai Chu, Brian Hon-Yin Chung","doi":"10.1093/bib/bbaf475","DOIUrl":"10.1093/bib/bbaf475","url":null,"abstract":"<p><p>Genomiser is a phenotype-driven tool that prioritizes coding and non-coding variants by relevance in rare disease diagnosis; yet comprehensive evaluation of its performance on real-life whole genome sequencing data is lacking. The Hong Kong Genome Project had initially incorporated Exomiser in the diagnostic pipeline. This study evaluated the feasibility of upgrading from Exomiser to Genomiser with three modifications: extension of the interval filter to include ±2000 bp from transcript boundaries, adjusting minor allele frequency (MAF) filter to 3%, and the inclusion of SpliceAI. A total of 985 patients with disclosed whole genome sequencing test results were included in this study, of which 207 positive cases (14 attributed to non-coding variants) were used for Genomiser parameter optimization by means of sensitivity evaluation. Under the default parameter setting, Genomiser achieved lower sensitivity compared to Exomiser (70.15% vs. 72.14%, top-3 candidates; 74.63% vs. 80.60%, top-5 candidates). Further investigation noted that this was attributed to non-coding variant noise influenced by Regulatory Mendelian Mutation (ReMM) scoring metrics. This issue was mitigated when a previously optimized ReMM score was applied as a filtering cut-off (ReMM = 0.963), improving Genomiser's sensitivity (92.54% vs. 89.55%, top-15 candidates). We further evaluated the optimized parameter in a cohort of 778 negative cases and detected 20 non-coding variants (2.6% added yield), with 5 validated to be disease-causing. Our proposed approach adheres to American College of Medical Genetics and Genomics/Association for Molecular Pathology and ClinGen variant interpretation guidelines to ensure interpretable results and integrates non-coding variant analysis into clinical pipelines.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12452281/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145124161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PEAKQC: periodicity evaluation in single-cell ATAC-seq data for quality assessment.","authors":"Jan Detleffsen, Brenton Bruns, Mette Bentsen, Carsten Kuenne, Mario Looso","doi":"10.1093/bib/bbaf465","DOIUrl":"10.1093/bib/bbaf465","url":null,"abstract":"<p><p>Chromatin organization guides gene regulatory mechanisms and has been subject of extensive research using chromatin accessibility assays. ATAC-seq is commonly applied to elucidate regulatory regions of the genome at both bulk and single-cell resolutions. However, the analysis of single-cell ATAC-seq data is particularly challenging due to issues such as data sparsity, low signal-to-noise ratios, and the lack of standardized quality control (QC) protocols. While QC based on the fragment length distribution (FLD) represents common practice for bulk analyses, an algorithmic solution that utilizes the full potential of the FLD at the single-cell level is missing. To address this limitation, we introduce the python package PEAKQC, a novel tool that provides a robust metric for identifying high-quality cells. PEAKQC quantifies the deviation of individual cells' FLD patterns from the expected distribution using a wavelet transformation-based convolution approach. Benchmarking against alternative metrics revealed favorable selection of high-quality cells, facilitating accurate downstream analysis including cell type identification and cluster separation. PEAKQC is readily installable via the Python Package Index and can be seamlessly integrated into existing single-cell analysis frameworks that utilize Python. By providing a robust and scalable solution for single-cell ATAC-seq QC, PEAKQC addresses a significant knowledge gap in the field and proposes FLD patterns as a novel standard for data quality assessment.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12448717/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145085116","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Neighborhood enrichment for the identification of antigen-specific T-cell receptors.","authors":"Kseniia R Lupyr, Pavel V Shelyakin, Konstantin A Sobyanin, Ruslan A Martynov, Vladimir S Popov, Sevastyan O Rabdano, Olga S Nikitina, Yurii G Yanushevich, Ilya A Kofiadi, Dmitry B Staroverov, Mikhail Shugay, Dmitriy M Chudakov, Olga V Britanova","doi":"10.1093/bib/bbaf495","DOIUrl":"10.1093/bib/bbaf495","url":null,"abstract":"<p><p>Understanding T-cell receptor (TCR) specificity is not only essential for fundamental research, but could open up novel avenues for diagnostics, cancer immunotherapy, and the targeted treatment of autoimmune diseases. The immune system responds to challenges through groups of T-cells with similar TCR sequences. In recent years, searching for TCRs with an enrichment of similar sequences - neighbors - in a TCR repertoire has become a standard procedure for antigen-specific TCR identification. This study provides a systematic comparison of computational algorithms-ALICE, TCRNET, GLIPH2, and tcrdist3-that leverage neighborhood enrichment for antigen-specific TCR identification. Using published murine datasets from Lymphocytic choriomeningitis virus (LCMV) infection and novel datasets from Sputnik V vaccination and Mycobacterium tuberculosis (Mtb) infection, we evaluated the performance of these algorithms. To facilitate reproducible analysis, we developed TCRgrapher, an R library that integrates these pipelines into a user-friendly framework. TCRgrapher enables efficient identification of antigen-specific TCRs from single repertoire snapshots and supports flexible parameter customization. Our comparative analysis revealed that ALICE and TCRNET consistently outperformed GLIPH2 and tcrdist3 across most datasets, achieving higher area under precision-recall curve. While murine datasets provide valuable insights into algorithm performance, caution is advised when extrapolating these results to other species or different experimental conditions. TCRgrapher is freely available on GitHub (https://github.com/KseniaMIPT/tcrgrapher), offering researchers a robust tool for investigating TCR specificity and advancing immunological studies.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12461701/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145136533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Significantly enhancing human antibody affinity via deep learning and computational biology-guided single-point mutations.","authors":"Junxin Li, Chao Zhang, Wei Xia, Hei Wun Kan, Kaifang Huang, Sai Li, Mark Akinola Ige, Qiuliyang Yu, Jiawei Zhao, Xiaochun Wan, John Z H Zhang, Haiping Zhang","doi":"10.1093/bib/bbaf445","DOIUrl":"10.1093/bib/bbaf445","url":null,"abstract":"<p><p>Enhancing antibody affinity is a critical goal in antibody design, as it improves therapeutic efficacy, specificity, and safety while reducing dosage requirements. Traditional methods, such as single-point mutations or combinatorial mutagenesis, are limited by the impracticality of exhaustively exploring the vast mutational space. To address this challenge, we developed a novel computational pipeline that integrates evolutionary constraints, antibody-antigen-specific statistical potentials, molecular dynamics simulations, metadynamics, and a suite of deep learning models to identify affinity-enhancing mutations. Our deep learning framework includes MicroMutate, which predicts microenvironment-specific amino acid mutations, and graph-based models that evaluate postmutation antigen-antibody-binding probabilities. Using this approach, we screened 12 single-point mutant antibodies targeting the hemagglutinin of the H7N9 avian influenza virus, starting from antibodies with initial affinities in the subnanomolar range, with one showing a 4.62-fold improvement. To demonstrate the generalizability of our method, we applied it to engineer an antibody against death receptor 5 with initial affinities in the subnanomolar range, successfully identifying a mutant with a 2.07-fold increase in affinity. Our work underscores the transformative potential of integrating deep learning and computational methods for rapidly and precisely discovering affinity-enhancing mutations while preserving immunogenicity and expression. This approach offers a powerful and universal platform for advancing antibody therapeutics.</p>","PeriodicalId":9209,"journal":{"name":"Briefings in bioinformatics","volume":"26 5","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12400800/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144943461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}