PolyConf: Unlocking Polymer Conformation Generation through Hierarchical Generative Models

Fanmeng Wang, Wentao Guo, Qi Ou, Hongshuai Wang, Haitao Lin, Hongteng Xu, Zhifeng Gao

TL;DR: This study introduces a pioneering method that leverages hierarchical generative models to unlock polymer conformation generation.

Abstract:

Polymer conformation generation is a critical task that enables atomic-level studies of diverse polymer materials.
While significant advances have been made in designing conformation generation methods for small molecules and proteins, these methods struggle to generate polymer conformations due to their unique structural characteristics.
Meanwhile, the scarcity of polymer conformation datasets further limits the progress, making this important area largely unexplored.
In this work, we propose PolyConf, a pioneering tailored polymer conformation generation method that leverages hierarchical generative models to unlock new possibilities.
Specifically, we decompose the polymer conformation into a series of local conformations (i.e., the conformations of its repeating units), generating these local conformations through an autoregressive model, and then generating their orientation transformations via a diffusion model to assemble them into the complete polymer conformation.
Moreover, we develop the first benchmark with a high-quality polymer conformation dataset derived from molecular dynamics simulations to boost related research in this area.
The comprehensive evaluation demonstrates that PolyConf consistently outperforms existing conformation generation methods, thus facilitating advancements in polymer modeling and simulation.
The whole work is available at https://polyconf-icml25.github.io.

Lay Summary:

In this work, we successfully unlock an important yet largely unexplored task in the context of machine learning—polymer conformation generation.
Specifically, we propose PolyConf, the first tailored polymer conformation generation method that leverages hierarchical generative models to tackle this task, and further develop PolyBench, the first benchmark for polymer conformation generation that comprises a high-quality polymer conformation dataset derived from molecular dynamics simulations to overcome the data scarcity.
Comprehensive experiments on the PolyBench benchmark demonstrate that our PolyConf significantly outperforms existing conformation generation methods in both quality and efficiency while maintaining superior scalability and generalization capabilities, thus facilitating advancements in polymer modeling and simulation.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Link To Code: https://polyconf-icml25.github.io

Primary Area: Applications->Chemistry, Physics, and Earth Sciences

Keywords: Polymer conformation generation, Polymer modeling, Hierarchical generative models

Submission Number: 2896