Join us on June 26th to hear from Riddhish Pandharkar, Scientific Developer II at OpenEye, Cadence Molecular Sciences, about Bayesian sampling in OMEGA for small molecule conformer generation.
About this session
This talk presents a new Bayesian approach to generating conformers in OMEGA, our small-molecule conformer generation tool. The new method is particularly well-suited for applications that require processing very large databases of molecules to generate a small number of conformations, e.g, 3D similarity searching with ROCS/FastROCS and structure-based machine learning. Instead of the core method of combinatorial sampling of torsions, which requires the generation of many candidate conformations for flexible molecules, Thompson sampling is used to dynamically bias sampling towards the low-strain conformers of the molecule.
Our results show a 1.5x-2x speed-up for generating FastROCS ensembles across various datasets, with the biggest improvements in speed seen for molecules with more than 6 rotors. We compare performance in reproducing bioactive conformations and ligand-based virtual screening using ROCS to show that the accuracy and quality of the ensemble are maintained despite this increase in speed.
About the presenter
This session will be presented by Riddhish Pandharkar PhD, Scientific Developer II at OpenEye, Cadence Molecular Sciences. Riddhish's work focuses on small-molecule modeling, particularly on the improved representation of torsions for conformer generation and force fields. Before joining OpenEye, he did his graduate studies at the University of Chicago, where he worked on quantum chemistry and electronic structure method development.
Webinar Details:
📣Title: Bayesian sampling in OMEGA for small molecule conformer generation.
📅 Date: June 26, 2025
🕛 Time: 11-11:30 am EDT