RL Boltzmann Generators for Conformer Generation in Data-Sparse Environments

TL;DR

This paper explores reinforcement learning Boltzmann generators for conformer generation in data-sparse environments like intrinsically disordered proteins, highlighting challenges in conformer coverage and training strategies.

Contribution

It introduces an RL-based Boltzmann generator trained on a Gibbs score, revealing limitations in conformer coverage and the independence of training efficacy from the modeling modality.

Findings

Training against the Gibbs score does not improve conformer coverage.

Energy-based training alone is insufficient for IDPs.

Mode collapse remains a challenge in conformer generation.

Abstract

The generation of conformers has been a long-standing interest to structural chemists and biologists alike. A subset of proteins known as intrinsically disordered proteins (IDPs) fail to exhibit a fixed structure and, therefore, must also be studied in this light of conformer generation. Unlike in the small molecule setting, ground truth data are sparse in the IDP setting, undermining many existing conformer generation methods that rely on such data for training. Boltzmann generators, trained solely on the energy function, serve as an alternative but display a mode collapse that similarly preclude their direct application to IDPs. We investigate the potential of training an RL Boltzmann generator against a closely related "Gibbs score," and demonstrate that conformer coverage does not track well with such training. This suggests that the inadequacy of solely training against the energy is independent of the modeling modality