Align Your Structures: Generating Trajectories with Structure Pretraining for Molecular Dynamics

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Iyengar, Aniketh, Han, Jiaqi, Sun, Pengwei, Jiang, Mingjian, Xie, Jianwen, Ermon, Stefano
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917384082685952
author Iyengar, Aniketh
Han, Jiaqi
Sun, Pengwei
Jiang, Mingjian
Xie, Jianwen
Ermon, Stefano
author_facet Iyengar, Aniketh
Han, Jiaqi
Sun, Pengwei
Jiang, Mingjian
Xie, Jianwen
Ermon, Stefano
contents Generating molecular dynamics (MD) trajectories using deep generative models has attracted increasing attention, yet remains inherently challenging due to the limited availability of MD data and the complexities involved in modeling high-dimensional MD distributions. To overcome these challenges, we propose a novel framework that leverages structure pretraining for MD trajectory generation. Specifically, we first train a diffusion-based structure generation model on a large-scale conformer dataset, on top of which we introduce an interpolator module trained on MD trajectory data, designed to enforce temporal consistency among generated structures. Our approach effectively harnesses abundant structural data to mitigate the scarcity of MD trajectory data and effectively decomposes the intricate MD modeling task into two manageable subproblems: structural generation and temporal alignment. We comprehensively evaluate our method on the QM9 and DRUGS small-molecule datasets across unconditional generation, forward simulation, and interpolation tasks, and further extend our framework and analysis to tetrapeptide and protein monomer systems. Experimental results confirm that our approach excels in generating chemically realistic MD trajectories, as evidenced by remarkable improvements of accuracy in geometric, dynamical, and energetic measurements.
format Preprint
id arxiv_https___arxiv_org_abs_2604_03911
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Align Your Structures: Generating Trajectories with Structure Pretraining for Molecular Dynamics
Iyengar, Aniketh
Han, Jiaqi
Sun, Pengwei
Jiang, Mingjian
Xie, Jianwen
Ermon, Stefano
Machine Learning
Quantitative Methods
Generating molecular dynamics (MD) trajectories using deep generative models has attracted increasing attention, yet remains inherently challenging due to the limited availability of MD data and the complexities involved in modeling high-dimensional MD distributions. To overcome these challenges, we propose a novel framework that leverages structure pretraining for MD trajectory generation. Specifically, we first train a diffusion-based structure generation model on a large-scale conformer dataset, on top of which we introduce an interpolator module trained on MD trajectory data, designed to enforce temporal consistency among generated structures. Our approach effectively harnesses abundant structural data to mitigate the scarcity of MD trajectory data and effectively decomposes the intricate MD modeling task into two manageable subproblems: structural generation and temporal alignment. We comprehensively evaluate our method on the QM9 and DRUGS small-molecule datasets across unconditional generation, forward simulation, and interpolation tasks, and further extend our framework and analysis to tetrapeptide and protein monomer systems. Experimental results confirm that our approach excels in generating chemically realistic MD trajectories, as evidenced by remarkable improvements of accuracy in geometric, dynamical, and energetic measurements.
title Align Your Structures: Generating Trajectories with Structure Pretraining for Molecular Dynamics
topic Machine Learning
Quantitative Methods
url https://arxiv.org/abs/2604.03911