Benchmarking Antimicrobial Peptide Identification with Sequence and Structure Representation
Keywords: Benchmark, AMP, drug discovery, multi-modal learning
Abstract: The rapid evolution of drug-resistant (DR) microbial has become a severe issue for human health. Antimicrobial peptides (AMPs) are powerful therapeutic drugs for treating DR microbial, but their clinical application is limited by activity and toxicity. Recently, AI has shown its power in discovering the high-activity AMPs, relying on the database of the AMP's wet-lab activity data. However, the activity data from this database are collected from thousands of papers, with their different wet lab experiments setting on one or few types of DR bacteria, have further limits the development of AI methods for AMP identification. Moreover, recently AlphaFold has revolutionized the field of drug discovery, but how can we benefit from the predicted structure for AMP discovery still remains unknown. To address the above challenges, we make two contributions. \textbf{a)} We construct the \textbf{DRAMPAtlas 1.0} that contains the training set collected from the public and the testing set from our wet lab experiment. Each AMP sequence is equipped with its 3D structure, activity data, and toxicity, where the activity is about six types of DR bacteria. \textbf{b)} We conduct extensive experiments for AMP identification, by modeling the 3D structure as voxels or graphs, in conjugate with its sequence information or solely with the structure or sequence. We have made many interesting findings. We hope that our benchmark and findings can benefit the research community to better design the algorithms for high-activity AMP discovery. All code and data associated with the work will be made publicly available after acceptance.
Primary Area: datasets and benchmarks
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 1303
Loading