Moltiverse: Molecular Conformer Generation Using Enhanced Sampling Methods

Accurately predicting the diverse bound-state conformations of small molecules is crucial for successful drug discovery, particularly when detailed protein-ligand interactions are unknown. Established tools exist, but exploring the vast conformational space efficiently remains a challenge. This work introduces Moltiverse, a novel protocol using enhanced sampling molecular dynamics (MD) simulations for conformer generation. The extended adaptive biasing force (eABF) algorithm combined with metadynamics, guided by a single collective variable (RDGYR), efficiently navigates the conformational landscape of each molecule. Benchmarked against established software like RDKit, CONFORGE, ConfGenX, Torsional diffusion, and Conformator, Moltiverse demonstrates comparable accuracy while achieving competitive performance in several test cases. Using eight quantitative metrics and statistical analysis, we present a systematic ranking of conformer generation algorithms and provide recommendations for their improvement based on our findings. We introduce the Cofactorv1 dataset, a complementary resource for conformer generator evaluation. Unlike traditional datasets with thousands of single-conformer molecules, Cofactorv1 dataset features only seven small molecule cofactors but with hundreds to thousands of experimental conformers per molecule (sourced from the PDB). This diversity, encompassing 15-29 rotatable bonds, poses a great challenge for conformer generation benchmarks. Cofactorv1 is a complementary dataset that serves as a valuable resource for developing and evaluating conformer generation methods like Moltiverse, pushing the boundaries of accuracy and diversity in this relevant field.