Backmapping coarse-grained models to all-atom configurations

Develop reliable methods to backmap coarse-grained molecular representations used in molecular dynamics simulations to full all-atom configurations, enabling reconstruction of atomistic structures from coarse-grained simulations.

Background

Coarse-grained (CG) models reduce computational cost by representing groups of atoms with fewer effective beads, enabling simulations of larger systems or longer timescales than are feasible with all-atom molecular dynamics. However, many analyses and downstream tasks still require atomistic detail, making reconstruction of full all-atom configurations from CG representations a critical step.

The paper highlights this backmapping step as an unresolved challenge. While preliminary approaches exist—such as autoencoder-based schemes where an encoder learns a CG model and a decoder backmaps to an ensemble’s average structure, and diffusion-based generative methods that attempt atom-resolution reconstruction—developing generally robust and accurate backmapping remains an open need for practical CG workflows.