Torsional-GFN: a conditional conformation generator for small molecules

2025-07-15

AI Summary

To address the low sampling efficiency and poor Boltzmann distribution fidelity in small-molecule conformational generation, this work introduces Torsional-GFN—the first generative flow network (GFlowNet) tailored for conformational sampling. It defines actions in torsional angle space and employs a conditional graph neural network to encode molecular topology and local geometry, enabling sequential, reversible conformation construction. Crucially, it is trained end-to-end solely on physics- or empirically derived reward functions, without explicit energy potentials or trajectory-level supervision. The model exhibits zero-shot generalization to unseen bond lengths and angles. On multiple benchmarks, it achieves significantly higher Boltzmann-weighted coverage than conventional methods (e.g., RDKit, ETKDG, ConfGF). Moreover, conformations generated by Torsional-GFN yield superior performance in downstream binding affinity prediction, demonstrating strong generalization across diverse protein–ligand complexes.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Generating stable molecular conformations is crucial in several drug discovery applications, such as estimating the binding affinity of a molecule to a target. Recently, generative machine learning methods have emerged as a promising, more efficient method than molecular dynamics for sampling of conformations from the Boltzmann distribution. In this paper, we introduce Torsional-GFN, a conditional GFlowNet specifically designed to sample conformations of molecules proportionally to their Boltzmann distribution, using only a reward function as training signal. Conditioned on a molecular graph and its local structure (bond lengths and angles), Torsional-GFN samples rotations of its torsion angles. Our results demonstrate that Torsional-GFN is able to sample conformations approximately proportional to the Boltzmann distribution for multiple molecules with a single model, and allows for zero-shot generalization to unseen bond lengths and angles coming from the MD simulations for such molecules. Our work presents a promising avenue for scaling the proposed approach to larger molecular systems, achieving zero-shot generalization to unseen molecules, and including the generation of the local structure into the GFlowNet model.

Problem

• Generating stable molecular conformations for drug discovery
• Sampling conformations from Boltzmann distribution efficiently
• Zero-shot generalization to unseen molecular structures

Innovation

• Conditional GFlowNet for molecular conformations
• Samples torsion angles using reward function
• Zero-shot generalization to unseen molecules