{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T06:05:19Z","timestamp":1784613919257,"version":"3.55.0"},"reference-count":27,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T00:00:00Z","timestamp":1784592000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100020896","name":"Zayed Bin Sultan Center for Health Sciences, United Arab Emirates University","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100020896","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100006013","name":"United Arab Emirates University","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100006013","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Bioinform."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Missense variants in breast cancer remain diagnostically challenging due to their functional diversity and complex genomic contexts. Conventional laboratory assays for evaluating pathogenicity are labor-intensive, costly, and often impractical for large-scale screening, creating a pressing need for accurate, scalable, and clinically interpretable computational approaches.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>In this study, we present a novel deep learning framework for predicting the pathogenicity of breast cancer missense variants, integrating comprehensive preprocessing, advanced imputation, rigorous model benchmarking, and explainability. Genetic variants were curated from multiple genomic databases, annotated using the Ensembl Variant Effect Predictor (VEP), and processed with Variational Autoencoders (VAE) for missing-value imputation. Seven deep learning models, MLP, CNN, DNN, RNN, LSTM, GRU, and Transformer, were trained and evaluated across 11 performance metrics. To quantify performance stability, each model was trained across five random seeds; mean AUC \u00b1 SD across seeds is reported as the primary performance estimate, with the best-seed run used only for LIME and PMI interpretability analyses. Recursive feature elimination, permutation importance (PMI), and Local Interpretable Model-Agnostic Explanations (LIME) were employed to enhance transparency. Statistical analyses, including Z-tests, ANOVA, and calibration assessments, validated performance consistency and inter-model differences.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>GRU achieved the highest internal AUC (0.9956 [95% CI 0.9936\u20130.9972]; mean across five seeds 0.9941 \u00b1 0.0011), with precision 0.9967 and calibration ECE 0.0095. Externally, LSTM led with AUC 0.9457, exceeding all eleven standalone predictors benchmarked on the same set. Models showed strong alignment with conservation signals such as phyloP470way and Eigen-PC scores. Notably, the pipeline provides performance metrics with 95% confidence intervals and incorporates case-level LIME visualizations for true positive, true negative, false positive, and false negative predictions, bolstering interpretability and clinical relevance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>This work delivers one of the most comprehensive evaluations of deep learning in breast cancer variant classification to date. By combining high-performance sequential models with interpretable AI tools, the proposed framework provides a reproducible, transparent benchmark for variant pathogenicity prediction and a foundation for future research use and translation in cancer genomics.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fbinf.2026.1844071","type":"journal-article","created":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T05:36:27Z","timestamp":1784612187000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Clinically interpretable deep learning for breast cancer missense variant pathogenicity prediction"],"prefix":"10.3389","volume":"6","author":[{"given":"Rahaf M.","family":"Ahmad","sequence":"first","affiliation":[{"name":"Department of Genetics and Genomics, College of Medical and Health Sciences, United Arab Emirates University","place":["Al-Ain, United Arab Emirates"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Noura","family":"AlDhaheri","sequence":"additional","affiliation":[{"name":"Department of Genetics and Genomics, College of Medical and Health Sciences, United Arab Emirates University","place":["Al-Ain, United Arab Emirates"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mohd Saberi","family":"Mohamad","sequence":"additional","affiliation":[{"name":"Department of Genetics and Genomics, College of Medical and Health Sciences, United Arab Emirates University","place":["Al-Ain, United Arab Emirates"]},{"name":"Centre for Advanced Analytics, CEO for Artificial Intelligence, Faculty of Engineering and Technology, Multimedia University","place":["Melaka, Malaysia"]},{"name":"Department of Biosystems Engineering, Faculty of Agricultural Technology, Universitas Brawijaya","place":["Malang, Indonesia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bassam R.","family":"Ali","sequence":"additional","affiliation":[{"name":"Department of Genetics and Genomics, College of Medical and Health Sciences, United Arab Emirates University","place":["Al-Ain, United Arab Emirates"]},{"name":"Zayed Center for Heath Sciences, College of Medicine and Health Sciences, United Arab Emirates University","place":["Al-Ain, United Arab Emirates"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,7,21]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"105689","DOI":"10.1016\/j.ijmedinf.2024.105689","article-title":"Decoding the Black box: explainable AI (XAI) for cancer diagnosis, prognosis, and treatment planning-A state-of-the art systematic review","volume":"193","author":"Abas Mohamed","year":"2025","journal-title":"Int. J. Med. Inf."},{"key":"B2","doi-asserted-by":"publisher","first-page":"200110","DOI":"10.1016\/j.sasc.2024.200110","article-title":"A hybrid machine learning model for classifying gene mutations in cancer using LSTM, BiLSTM, CNN, GRU, and GloVe","volume":"6","author":"Aburass","year":"2024","journal-title":"Syst. Soft Comput."},{"key":"B3","doi-asserted-by":"publisher","first-page":"bbad479","DOI":"10.1093\/bib\/bbad479","article-title":"A review of genetic variant databases and machine learning tools for predicting the pathogenicity of breast cancer","volume":"25","author":"Ahmad","year":"2024","journal-title":"Briefings Bioinforma"},{"key":"B4","doi-asserted-by":"publisher","first-page":"99","DOI":"10.1186\/s40246-024-00667-9","article-title":"AI-derived comparative assessment of the performance of pathogenicity prediction tools on missense variants of breast cancer genes","volume":"18","author":"Ahmad","year":"2024","journal-title":"Hum. Genomics"},{"key":"B5","doi-asserted-by":"publisher","first-page":"4668","DOI":"10.1016\/j.csbj.2025.10.052","article-title":"Leveraging AutoML to optimize dataset selection for improved breast cancer variant pathogenicity prediction","volume":"27","author":"Ahmad","year":"2025","journal-title":"Comput. Struct. Biotechnol. J"},{"key":"B6","first-page":"771","article-title":"A comparison of LSTM and GRU networks for learning symbolic sequences","volume-title":"Lecture Notes in Networks and Systems","author":"Cahuantzi","year":"2023"},{"key":"B7","doi-asserted-by":"publisher","first-page":"51","DOI":"10.1186\/s13073-022-01052-8","article-title":"Breast cancer risks associated with missense variants in breast cancer susceptibility genes","volume":"14","author":"Dorling","year":"2022","journal-title":"Genome Med."},{"key":"B8","doi-asserted-by":"publisher","first-page":"163","DOI":"10.2478\/acss-2020-0018","article-title":"Breast cancer prediction using stacked GRU-LSTM-BRNN","volume":"25","author":"Dutta","year":"2020","journal-title":"Appl. Comput. Syst."},{"key":"B9","doi-asserted-by":"publisher","first-page":"e32279","DOI":"10.1016\/j.heliyon.2024.e32279","article-title":"Deep-GenMut: automated genetic mutation classification in oncology: a deep learning comparative study","volume":"10","author":"Elsamahy","year":"2024","journal-title":"Heliyon"},{"key":"B10","doi-asserted-by":"publisher","first-page":"10478","DOI":"10.1038\/s41598-023-37698-6","article-title":"Gene-specific machine learning for pathogenicity prediction of rare BRCA1 and BRCA2 missense variants","volume":"13","author":"Kang","year":"2023","journal-title":"Sci. Rep."},{"key":"B11","doi-asserted-by":"publisher","first-page":"315","DOI":"10.1152\/physiolgenomics.00033.2023","article-title":"BRCA1-specific machine learning model predicts variant pathogenicity with high accuracy","volume":"55","author":"Khandakji","year":"2023","journal-title":"Physiol. Genomics"},{"key":"B12","doi-asserted-by":"publisher","first-page":"311","DOI":"10.3390\/jimaging10120311","article-title":"State-of-the-Art deep learning methods for microscopic image segmentation: applications to cells, nuclei, and tissues","volume":"10","author":"Krikid","year":"2024","journal-title":"J. Imaging"},{"key":"B13","doi-asserted-by":"publisher","first-page":"e1002201","DOI":"10.1371\/journal.pmed.1002201","article-title":"Mutational profile of metastatic breast cancers: a retrospective analysis","volume":"13","author":"Lefebvre","year":"2016","journal-title":"PLoS Med."},{"key":"B14","doi-asserted-by":"publisher","first-page":"22823","DOI":"10.1609\/aaai.v40i27.39445","article-title":"Tab-PET: graph-based positional encodings for tabular transformers","volume":"40","author":"Leng","year":"2026","journal-title":"Proc. AAAI Conf. Artif. Intell."},{"key":"B15","doi-asserted-by":"crossref","DOI":"10.3389\/fdata.2025.1515341","volume-title":"A Review of AI-based Radiogenomics in Neurodegenerative Disease","author":"Liu","year":"2025"},{"key":"B16","doi-asserted-by":"publisher","first-page":"122","DOI":"10.1186\/s13059-016-0974-4","article-title":"The ensembl variant effect predictor","volume":"17","author":"McLaren","year":"2016","journal-title":"Genome Biol."},{"key":"B17","doi-asserted-by":"publisher","first-page":"14604582241307839","DOI":"10.1177\/14604582241307839","article-title":"Researching public health datasets in the era of deep learning: a systematic literature review","volume":"31","author":"Obeidat","year":"2025","journal-title":"Health Inf. J."},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.1093\/gigascience\/giaa082","article-title":"Genomic data imputation with variational auto-encoders","volume":"9","author":"Qiu","year":"2020","journal-title":"Gigascience"},{"key":"B19","volume-title":"Evaluation of GPT and BERT-based Models on Identifying protein-protein Interactions in Biomedical Text","author":"Rehana","year":"2023"},{"key":"B20","doi-asserted-by":"publisher","first-page":"405","DOI":"10.1038\/gim.2015.30","article-title":"Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American college of medical genetics and genomics and the association for molecular pathology","volume":"17","author":"Richards","year":"2015","journal-title":"Genet. Med."},{"key":"B21","doi-asserted-by":"crossref","DOI":"10.1101\/2025.02.13.25322245","volume-title":"Putative Breast Cancer Risk Variants from Populations of South Asian Ancestry Are Under-represented in Public Variant Classification Databases","author":"Rony","year":"2025"},{"key":"B22","volume-title":"Machine Learning and Prediction of Infectious Diseases: A Systematic Review","author":"Santangelo","year":"2023"},{"key":"B23","doi-asserted-by":"publisher","first-page":"e2444","DOI":"10.7717\/peerj-cs.2444","article-title":"Hybrid transformer-CNN and LSTM model for lung disease segmentation and classification","volume":"10","author":"Shafi","year":"2024","journal-title":"PeerJ Comput. Sci."},{"key":"B24","article-title":"A comprehensive overview and comparative analysis on deep learning models: CNN, RNN, LSTM, GRU","author":"Shiri","year":"2023"},{"key":"B25","volume-title":"High-Performance Medicine: The Convergence of Human and Artificial Intelligence","author":"Topol","year":"2019"},{"key":"B26","first-page":"5999","article-title":"Attention is all you need","volume-title":"Advances in Neural Information Processing Systems, Neural Information Processing Systems Foundation","author":"Vaswani","year":"2017"},{"key":"B27","volume-title":"A Comparison Review of Transfer Learning and self-supervised Learning: Definitions, Applications, Advantages and Limitations","author":"Zhao","year":"2024"}],"container-title":["Frontiers in Bioinformatics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fbinf.2026.1844071\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T05:36:30Z","timestamp":1784612190000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fbinf.2026.1844071\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,21]]},"references-count":27,"alternative-id":["10.3389\/fbinf.2026.1844071"],"URL":"https:\/\/doi.org\/10.3389\/fbinf.2026.1844071","relation":{},"ISSN":["2673-7647"],"issn-type":[{"value":"2673-7647","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,21]]},"article-number":"1844071"}}