{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,10,2]],"date-time":"2026-10-02T02:54:11Z","timestamp":1790909651095,"version":"4.1.0"},"posted":{"date-parts":[[2026,9,29]]},"group-title":"Genomics","reference-count":70,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T00:00:00Z","timestamp":1790640000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"award":["R01AG06179"],"award-info":[{"award-number":["R01AG06179"]}],"id":[{"id":"https:\/\/ror.org\/001d55x84","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,9,29]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:p>Recent deep learning (DL) models predict average gene expression levels from DNA sequences with high overall correlation to measured values. We examine these DL models through the lens of disease research. The seemingly high overall performance of DL models is largely due to capturing whether genes are \u201cOn\u201d or \u201cOff\u201d, and to a lesser extent, disease-relevant expression level changes for \u201cOn\u201d genes. Indeed, the more bimodal the gene expression distribution, the better the reported performance. We track the extent of this issue across tissues, molecular systems, and cancers. Compounding the model evaluation issue, we find inconsistencies between the published code and reported performance, highlighting the importance of versioning and publishing performance evaluation code. These findings indicate that the high reported performance of popular DL models falls unexpectedly short in disease applications, and that the problem of personalized genomic prediction remains far from solved in a disease context.<\/jats:p>","DOI":"10.64898\/2026.09.24.753534","type":"posted-content","created":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T05:00:17Z","timestamp":1790744417000},"source":"Crossref","is-referenced-by-count":0,"title":["Disease relevance and replicability of deep learning gene expression prediction"],"prefix":"10.64898","author":[{"given":"Ada","family":"Zhang","sequence":"first","affiliation":[{"name":"Department of Psychiatry, SUNY Upstate Medical University, Syracuse, NY, USA"},{"name":"Rush Alzheimer\u2019s Disease Center, Rush University Medical Center, Chicago, IL, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3656-7394","authenticated-orcid":false,"given":"Shinya","family":"Tasaki","sequence":"additional","affiliation":[{"name":"Rush Alzheimer\u2019s Disease Center, Rush University Medical Center, Chicago, IL, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"Connell","sequence":"additional","affiliation":[{"name":"Rush Alzheimer\u2019s Disease Center, Rush University Medical Center, Chicago, IL, USA"},{"name":"Department of Neurology, Columbia University Irving Medical Center, New York, NY, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bernard","family":"Ng","sequence":"additional","affiliation":[{"name":"Department of Psychiatry, SUNY Upstate Medical University, Syracuse, NY, USA"},{"name":"Rush Alzheimer\u2019s Disease Center, Rush University Medical Center, Chicago, IL, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chris","family":"Gaiteri","sequence":"additional","affiliation":[{"name":"Department of Psychiatry, SUNY Upstate Medical University, Syracuse, NY, USA"},{"name":"Rush Alzheimer\u2019s Disease Center, Rush University Medical Center, Chicago, IL, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026100119251631000_2026.09.24.753534v1.1","doi-asserted-by":"publisher","DOI":"10.1038\/s41588-021-00921-z"},{"key":"2026100119251631000_2026.09.24.753534v1.2","doi-asserted-by":"publisher","DOI":"10.1161\/CIRCRESAHA.117.312086"},{"key":"2026100119251631000_2026.09.24.753534v1.3","doi-asserted-by":"publisher","DOI":"10.1016\/j.cell.2024.12.002"},{"key":"2026100119251631000_2026.09.24.753534v1.4","doi-asserted-by":"publisher","DOI":"10.1038\/s41588-020-0637-y"},{"key":"2026100119251631000_2026.09.24.753534v1.5","doi-asserted-by":"crossref","first-page":"1685","DOI":"10.1038\/s41380-018-0059-8","article-title":"Differential activity of transcribed enhancers in the prefrontal cortex of 537 cases with schizophrenia and controls","volume":"24","year":"2019","journal-title":"Molecular psychiatry"},{"key":"2026100119251631000_2026.09.24.753534v1.6","doi-asserted-by":"publisher","DOI":"10.1016\/j.cell.2016.03.008"},{"key":"2026100119251631000_2026.09.24.753534v1.7","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pgen.1000888"},{"key":"2026100119251631000_2026.09.24.753534v1.8","doi-asserted-by":"publisher","DOI":"10.1038\/ng1589"},{"key":"2026100119251631000_2026.09.24.753534v1.9","doi-asserted-by":"crossref","first-page":"e1203","DOI":"10.1038\/tp.2017.154","article-title":"Rna-seq analysis of amygdala tissue reveals characteristic expression profiles in schizophrenia","volume":"7","year":"2017","journal-title":"Translational psychiatry"},{"key":"2026100119251631000_2026.09.24.753534v1.10","doi-asserted-by":"publisher","DOI":"10.1161\/01.RES.0000159390.03503.c3"},{"key":"2026100119251631000_2026.09.24.753534v1.11","doi-asserted-by":"publisher","DOI":"10.1038\/S41593-020-0621-Y"},{"key":"2026100119251631000_2026.09.24.753534v1.12","doi-asserted-by":"crossref","first-page":"e70025","DOI":"10.1002\/alz.70025","article-title":"Systematic review and meta-analysis of bulk rnaseq studies in human alzheimer\u2019s disease brain tissue","volume":"21","year":"2025","journal-title":"Alzheimer\u2019s & Dementia"},{"key":"2026100119251631000_2026.09.24.753534v1.13","doi-asserted-by":"publisher","DOI":"10.1158\/0008-5472.CAN-15-0484"},{"key":"2026100119251631000_2026.09.24.753534v1.14","doi-asserted-by":"publisher","DOI":"10.1038\/s41588-020-00726-6"},{"key":"2026100119251631000_2026.09.24.753534v1.15","doi-asserted-by":"crossref","first-page":"24","DOI":"10.1016\/j.cels.2019.06.006","article-title":"Before and after: comparison of legacy and harmonized tcga genomic data commons\u2019 data","volume":"9","year":"2019","journal-title":"Cell systems"},{"key":"2026100119251631000_2026.09.24.753534v1.16","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/gkw1064"},{"key":"2026100119251631000_2026.09.24.753534v1.17","doi-asserted-by":"crossref","first-page":"931","DOI":"10.1038\/nmeth.3547","article-title":"Predicting effects of noncoding variants with deep learning\u2013based sequence model","volume":"12","year":"2015","journal-title":"Nature methods"},{"key":"2026100119251631000_2026.09.24.753534v1.18","doi-asserted-by":"publisher","DOI":"10.1126\/science.1242088"},{"key":"2026100119251631000_2026.09.24.753534v1.19","doi-asserted-by":"publisher","DOI":"10.1126\/science.1230062"},{"key":"2026100119251631000_2026.09.24.753534v1.20","doi-asserted-by":"publisher","DOI":"10.1038\/nature09266"},{"key":"2026100119251631000_2026.09.24.753534v1.21","doi-asserted-by":"publisher","DOI":"10.1038\/nm.3384"},{"key":"2026100119251631000_2026.09.24.753534v1.22","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-024-07316-0"},{"key":"2026100119251631000_2026.09.24.753534v1.23","doi-asserted-by":"publisher","DOI":"10.1038\/ng.3314"},{"key":"2026100119251631000_2026.09.24.753534v1.24","article-title":"Long context rag performance of large language models","year":"2024","journal-title":"arXiv preprint"},{"key":"2026100119251631000_2026.09.24.753534v1.25","unstructured":"Shi, J. et al. Explaining context length scaling and bounds for language models. arXiv e-prints arXiv\u20132502 (2025)."},{"key":"2026100119251631000_2026.09.24.753534v1.26","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-025-10014-0"},{"key":"2026100119251631000_2026.09.24.753534v1.27","doi-asserted-by":"publisher","DOI":"10.1038\/s41592-021-01252-x"},{"key":"2026100119251631000_2026.09.24.753534v1.28","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pcbi"},{"key":"2026100119251631000_2026.09.24.753534v1.29","doi-asserted-by":"publisher","DOI":"10.1038\/s41588-023-01524-6"},{"key":"2026100119251631000_2026.09.24.753534v1.30","doi-asserted-by":"publisher","DOI":"10.1038\/s41588-023-01574-w"},{"key":"2026100119251631000_2026.09.24.753534v1.31","doi-asserted-by":"crossref","unstructured":"Li, S. , Luo, R. & Huang, Y. Assessing large-scale genomic language models in predicting personal gene expression: promises and limitations. bioRxiv 2025\u201307 (2025).","DOI":"10.1101\/2025.07.09.664024"},{"key":"2026100119251631000_2026.09.24.753534v1.32","doi-asserted-by":"crossref","unstructured":"Drusinsky, S. , Whalen, S. & Pollard, K. S. Deep-learning prediction of gene expression from personal genomes. Genome Biology (2026).","DOI":"10.1186\/s13059-025-03926-7"},{"key":"2026100119251631000_2026.09.24.753534v1.33","doi-asserted-by":"crossref","unstructured":"Rastogi, R. , Reddy, A. J. , Chung, R. & Ioannidis, N. M. Fine-tuning sequence-to-expression models on personal genome and transcriptome data. bioRxiv 2024\u201309 (2024).","DOI":"10.1101\/2024.09.23.614632"},{"key":"2026100119251631000_2026.09.24.753534v1.34","doi-asserted-by":"crossref","unstructured":"Vaz, E. , Wang, L. , Galvin, J. , Keener, R. & Battle, A. Fine-tuning sequence to function deep learning models on large-scale proteomic data improves the accuracy of variant effect prediction. bioRxiv (2025).","DOI":"10.1101\/2025.09.26.678908"},{"key":"2026100119251631000_2026.09.24.753534v1.35","doi-asserted-by":"crossref","unstructured":"Spiro, A. E. et al. A scalable approach to investigating sequence-to-function predictions from personal genomes. bioRxiv 2025\u201302 (2025).","DOI":"10.1101\/2025.02.21.639494"},{"key":"2026100119251631000_2026.09.24.753534v1.36","doi-asserted-by":"crossref","unstructured":"Barbadilla-Mart\u00ednez, L. , Klaassen, N. , van Steensel, B. & de Ridder, J. Predicting gene expression from DNA sequence using deep learning models. Nat. Rev. Genet. 1\u201315 (2025).","DOI":"10.1038\/s41576-025-00841-2"},{"key":"2026100119251631000_2026.09.24.753534v1.37","doi-asserted-by":"publisher","DOI":"10.1126\/science.aat8127"},{"key":"2026100119251631000_2026.09.24.753534v1.38","doi-asserted-by":"publisher","DOI":"10.1038\/s41593-018-0154-9"},{"key":"2026100119251631000_2026.09.24.753534v1.39","doi-asserted-by":"publisher","DOI":"10.1038\/nn.4399"},{"key":"2026100119251631000_2026.09.24.753534v1.40","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1804106115"},{"key":"2026100119251631000_2026.09.24.753534v1.41","doi-asserted-by":"publisher","DOI":"10.1126\/science.1249168"},{"key":"2026100119251631000_2026.09.24.753534v1.42","doi-asserted-by":"publisher","DOI":"10.1126\/science.ade4401"},{"key":"2026100119251631000_2026.09.24.753534v1.43","doi-asserted-by":"publisher","DOI":"10.1016\/0092-8674(91)90418-X"},{"key":"2026100119251631000_2026.09.24.753534v1.44","doi-asserted-by":"crossref","unstructured":"Ma\u00dfberg, D. & Hatt, H. Human olfactory receptors: novel cellular functions outside of the nose. Physiological reviews (2018).","DOI":"10.1152\/physrev.00013.2017"},{"key":"2026100119251631000_2026.09.24.753534v1.45","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbc.2021.100475"},{"key":"2026100119251631000_2026.09.24.753534v1.46","doi-asserted-by":"publisher","DOI":"10.1126\/science.1080376"},{"key":"2026100119251631000_2026.09.24.753534v1.47","doi-asserted-by":"crossref","first-page":"13978","DOI":"10.1038\/s41598-017-14086-5","article-title":"Activation of intestinal olfactory receptor stimulates glucagon-like peptide-1 secretion in enteroendocrine cells and attenuates hyperglycemia in type 2 diabetic mice","volume":"7","year":"2017","journal-title":"Scientific Reports"},{"key":"2026100119251631000_2026.09.24.753534v1.48","doi-asserted-by":"publisher","DOI":"10.1074\/jbc.M109.012096"},{"key":"2026100119251631000_2026.09.24.753534v1.49","doi-asserted-by":"publisher","DOI":"10.1016\/j.devcel.2009.09.004"},{"key":"2026100119251631000_2026.09.24.753534v1.50","doi-asserted-by":"publisher","DOI":"10.1126\/science.abg3067"},{"key":"2026100119251631000_2026.09.24.753534v1.51","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.0307882100"},{"key":"2026100119251631000_2026.09.24.753534v1.52","doi-asserted-by":"publisher","DOI":"10.1186\/gb-2001-2-7-reports0018"},{"key":"2026100119251631000_2026.09.24.753534v1.53","doi-asserted-by":"publisher","DOI":"10.1101\/gr.5185406"},{"key":"2026100119251631000_2026.09.24.753534v1.54","doi-asserted-by":"publisher","DOI":"10.1101\/gr.126201.111"},{"key":"2026100119251631000_2026.09.24.753534v1.55","doi-asserted-by":"publisher","DOI":"10.1038\/ng.3141"},{"key":"2026100119251631000_2026.09.24.753534v1.56","doi-asserted-by":"publisher","DOI":"10.1038\/ng.3101"},{"key":"2026100119251631000_2026.09.24.753534v1.57","doi-asserted-by":"publisher","DOI":"10.1016\/j.cell.2013.09.053"},{"key":"2026100119251631000_2026.09.24.753534v1.58","doi-asserted-by":"crossref","first-page":"5958","DOI":"10.1038\/s41467-023-41690-z","article-title":"Genome-wide enhancer-gene regulatory maps link causal variants to target genes underlying human cancer risk","volume":"14","year":"2023","journal-title":"Nature communications"},{"key":"2026100119251631000_2026.09.24.753534v1.59","doi-asserted-by":"crossref","first-page":"158","DOI":"10.1186\/s13073-021-00970-3","article-title":"Integrative epigenomic and high-throughput functional enhancer profiling reveals determinants of enhancer heterogeneity in gastric cancer","volume":"13","year":"2021","journal-title":"Genome medicine"},{"key":"2026100119251631000_2026.09.24.753534v1.60","doi-asserted-by":"publisher","DOI":"10.1186\/s13059-015-0668-3"},{"key":"2026100119251631000_2026.09.24.753534v1.61","doi-asserted-by":"crossref","unstructured":"Lin, J. , Li, Z. , Zhao, Y. , Luo, R. & Pinello, L. Epinformer: scalable and integrative prediction of gene expression from promoter-enhancer sequences with multimodal epigenomic profiles. Nature Communications (2026).","DOI":"10.1038\/s41467-026-70535-8"},{"key":"2026100119251631000_2026.09.24.753534v1.62","first-page":"1","article-title":"Olfactory receptors and human diseases","volume":"401","year":"2025","journal-title":"Cell and Tissue Research"},{"key":"2026100119251631000_2026.09.24.753534v1.63","doi-asserted-by":"publisher","DOI":"10.1111\/j.2517-6161.1951.tb00088.x"},{"key":"2026100119251631000_2026.09.24.753534v1.64","doi-asserted-by":"crossref","unstructured":"Mostafavi, S. et al. A scalable approach to investigating sequence-to-expression prediction from personal genomes (2025).","DOI":"10.21203\/rs.3.rs-6030940\/v1"},{"key":"2026100119251631000_2026.09.24.753534v1.65","unstructured":"Biewald, L. Experiment tracking with weights & biases (2020). URL https:\/\/www.wandb.com\/. Software available from wandb.com."},{"key":"2026100119251631000_2026.09.24.753534v1.66","first-page":"39","article-title":"Accelerating the machine learning lifecycle with MLflow","volume":"41","year":"2018","journal-title":"IEEE Data Eng. Bull"},{"key":"2026100119251631000_2026.09.24.753534v1.67","first-page":"1","article-title":"The fair guiding principles for scientific data management and stewardship","volume":"3","year":"2016","journal-title":"Scientific data"},{"key":"2026100119251631000_2026.09.24.753534v1.68","doi-asserted-by":"crossref","first-page":"1132","DOI":"10.1038\/s41592-021-01256-7","article-title":"Reproducibility standards for machine learning in the life sciences","volume":"18","year":"2021","journal-title":"Nature methods"},{"key":"2026100119251631000_2026.09.24.753534v1.69","doi-asserted-by":"crossref","first-page":"e70002","DOI":"10.1002\/aaai.70002","article-title":"Reproducibility in machine-learning-based research: Overview, barriers, and drivers","volume":"46","year":"2025","journal-title":"AI Magazine"},{"key":"2026100119251631000_2026.09.24.753534v1.70","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/gkae235"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.09.24.753534","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,2]],"date-time":"2026-10-02T02:25:23Z","timestamp":1790907923000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.64898\/2026.09.24.753534"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,29]]},"references-count":70,"URL":"https:\/\/doi.org\/10.64898\/2026.09.24.753534","relation":{},"subject":[],"published":{"date-parts":[[2026,9,29]]},"subtype":"preprint"}}