Active Learning on Synthons for Molecular Design
This addresses the problem of efficient molecular design for drug discovery, offering a scalable solution for combinatorial spaces, though it appears incremental as it extends existing active learning methods.
The paper tackles the intractability of exhaustive virtual screening in drug discovery, especially for ultra-large combinatorial molecular spaces, by introducing SALSA, an active learning algorithm that scales to trillions of compounds and achieves comparable chemical properties to known bioactives with greater diversity and higher scores than an industry-leading generative approach.
Exhaustive virtual screening is highly informative but often intractable against the expensive objective functions involved in modern drug discovery. This problem is exacerbated in combinatorial contexts such as multi-vector expansion, where molecular spaces can quickly become ultra-large. Here, we introduce Scalable Active Learning via Synthon Acquisition (SALSA): a simple algorithm applicable to multi-vector expansion which extends pool-based active learning to non-enumerable spaces by factoring modeling and acquisition over synthon or fragment choices. Through experiments on ligand- and structure-based objectives, we highlight SALSA's sample efficiency, and its ability to scale to spaces of trillions of compounds. Further, we demonstrate application toward multi-parameter objective design tasks on three protein targets - finding SALSA-generated molecules have comparable chemical property profiles to known bioactives, and exhibit greater diversity and higher scores over an industry-leading generative approach.