Recombination Flow Matching Model for Protein Evolution
Keywords: Flow matching, Generative model, Protein design
Abstract: The design of novel proteins, distinct from those found in nature, holds immense potential for advancing drug discovery, biotechnology, and material science. However, current methodologies often face significant limitations in generating both novel protein structures. Biological evolution, a natural process that fosters novelty, heavily relies on recombination—yet this mechanism remains largely untapped in protein design. In this work, we propose Recombination Flow Matching (RFM), a novel generative model inspired by the principles of evolution. RFM meticulously preserves the structural integrity of protein segments during recombination while autonomously optimizing their spatial arrangement within the resultant protein. Using a common benchmark dataset, we demonstrate that RFM significantly outperforms established methods in producing structurally novel proteins. This approach opens new frontiers in protein design, leveraging evolutionary recombination to enhance the novelty of protein design. To the best of our knowledge, RFM is the first model to incorporate recombination into protein design.
Primary Area: generative models
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.
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: 11433
Loading