MolStitch: Offline Multi-Objective Molecular Optimization with Molecular Stitching
Keywords: molecular optimization, offline optimization, drug discovery
TL;DR: A novel framework that stitches molecules from an offline dataset to fine-tune the generative model.
Abstract: Molecular discovery is essential for advancing various scientific fields by generating novel molecules with desirable properties. This process is naturally a multi-objective optimization problem, as it must balance multiple molecular properties simultaneously. Although numerous methods have been developed to address this problem, most rely on online settings that repeatedly evaluate candidate molecules through oracle queries. However, in practical applications, online settings may not be feasible due to the extensive time and resources required for each oracle query. To fill this gap, we propose the Molecular Stitching (MolStitch) framework, which utilizes a fixed offline dataset to explore and optimize molecules without the need for repeated oracle queries. Specifically, MolStitch leverages existing molecules from the offline dataset to generate novel `stitched molecules' that combine their desirable properties. These stitched molecules are then used as training samples to fine-tune the generative model, enhancing its ability to produce superior molecules beyond those in the offline dataset. Experimental results on various offline multi-objective molecular optimization problems validate the effectiveness of MolStitch. MolStitch has been thoroughly analyzed, and its source code is available online.
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
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: 6519
Loading