PURE: policy-guided unbiased REpresentations for structure-constrained molecular generation
Anju Gupta, Barathi Lenin, Sean F. Current, Rohit Batra, Balaraman Ravindran, Karthik Raman, Srinivasan Parthasarathy · Journal of Cheminformatics · 2025
Structure-constrained molecular generation (SCMG) generates novel molecules that are structurally similar to a given molecule and have optimized properties. Deep learning solutions for SCMG are limited in that they are predisposed towards existing knowledge, and they suffer from a natural impedance mismatch problem due to the discrete nature of molecules, while deep learning methods for SCMG often operate in continuous space. Moreover, many task-specific evaluation metrics used during training often bias the model towards a particular metric -“metric-leakage”. To overcome these shortcomings, we propose Policy-guided Unbiased REpresentations (PURE) for SCMG that learns within a framework simulating molecular transformations for drug synthesis. PURE combines self-supervised learning with a policy-based reinforcement learning (RL) framework, thereby avoiding the need for external molecular metrics while learning high-quality representations that incorporate an inherent notion of similarity specific to the given task. Along with a semi-supervised training design, PURE utilizes template-based molecular simulations to better explore and navigate the discrete molecular search space. Despite the lack of metric biases, PURE achieves competitive or superior performance to state-of-the-art methods on multiple benchmarks. Our study emphasizes the importance of reevaluating current approaches for SCMG and developing strategies that naturally align with the problem. Finally, we illustrate how our methodology can be applied to combat drug resistance by identifying sorafenib-like compounds as a case study. PURE is a novel approach to structure-constrained molecular generation that utilizes policy-guided training to reduce overfitting to task-specific metrics for SCMG models. This approach offers an efficient and unique method for generating new and diverse molecule structures similar to target molecules without relying on user-defined similarity metrics. We show that PURE is capable of achieving competitive or superior performance on multiple benchmark tasks compared to other state-of-the-art methods and generates multiple viable alternative molecules to reduce drug resistance in a case study.