Researchers from IIT Madras’ Wadhwani School of Data Science and Artificial Intelligence (WSAI) and The Ohio State University, U.S., have developed a breakthrough Artificial Intelligence (AI) framework that can rapidly generate drug-like molecules optimized for real-world synthesis.
The new system, named PURE — Policy-guided Unbiased REpresentations for Structure-Constrained Molecular Generation (SCMG) — has the potential to drastically shorten the early stages of drug development, a process that currently spans over a decade and costs billions of dollars. It could also play a pivotal role in combating drug resistance in cancer and infectious diseases.
Unlike existing AI tools that rely on rigid scoring systems or statistical optimization, PURE simulates the step-by-step process of molecular synthesis based on real chemical reactions. By combining self-supervised learning and policy-based reinforcement learning, the framework explores chemical possibilities more organically — without depending on external evaluation metrics. This approach reduces bias and enables the model to independently learn molecular similarities.
PURE was tested on widely recognized molecule-generation benchmarks such as QED (drug-likeness), DRD2 (dopamine receptor activity), and solubility. It outperformed other models by generating a higher diversity and novelty of molecules and even suggesting potential synthetic routes — despite not being trained on those specific metrics.
The study, published in the Journal of Cheminformatics (DOI: 10.1186/s13321-025-01090-5), marks a significant step forward in AI-driven molecular discovery.
The research team includes Abhor Gupta, Barathi Lenin, Rohit Batra, Prof. B. Ravindran, and Prof. Karthik Raman from the Robert Bosch Centre for Data Science and AI and Wadhwani School of Data Science and AI, IIT Madras, along with Prof. Srinivasan Parthasarathy and Sean Current from the Department of Computer Science and Engineering, The Ohio State University, U.S.