{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deepdiff-deep-learning-for-predicting","title":"DeepDiff: Deep-learning for predicting Differential gene expression from histone modifications","arxiv_id":"1807.03878","date":"2018-07-10","proceeding":null,"authors":["Arshdeep Sekhon","Ritambhara Singh","Yanjun Qi"],"abstract":"Computational methods that predict differential gene expression from histone\nmodification signals are highly desirable for understanding how histone\nmodifications control the functional heterogeneity of cells through influencing\ndifferential gene regulation. Recent studies either failed to capture\ncombinatorial effects on differential prediction or primarily only focused on\ncell type-specific analysis. In this paper, we develop a novel attention-based\ndeep learning architecture, DeepDiff, that provides a unified and end-to-end\nsolution to model and to interpret how dependencies among histone modifications\ncontrol the differential patterns of gene regulation. DeepDiff uses a hierarchy\nof multiple Long short-term memory (LSTM) modules to encode the spatial\nstructure of input signals and to model how various histone modifications\ncooperate automatically. We introduce and train two levels of attention jointly\nwith the target prediction, enabling DeepDiff to attend differentially to\nrelevant modifications and to locate important genome positions for each\nmodification. Additionally, DeepDiff introduces a novel deep-learning based\nmulti-task formulation to use the cell-type-specific gene expression\npredictions as auxiliary tasks, encouraging richer feature embeddings in our\nprimary task of differential expression prediction. Using data from Roadmap\nEpigenomics Project (REMC) for ten different pairs of cell types, we show that\nDeepDiff significantly outperforms the state-of-the-art baselines for\ndifferential gene expression prediction. The learned attention weights are\nvalidated by observations from previous studies about how epigenetic mechanisms\nconnect to differential gene expression. Codes and results are available at\n\\url{deepchrome.org}","url_abs":"http://arxiv.org/abs/1807.03878v1","url_pdf":"http://arxiv.org/pdf/1807.03878v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deepdiff-deep-learning-for-predicting","repo_url":"https://github.com/QData/DeepDiffChrome","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"prediction","task_name":"Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}