Protein Graph Neural Networks for Heterogeneous Cryo-EM Reconstruction

TL;DR

This paper introduces a geometry-aware graph neural network approach for heterogeneous cryo-EM reconstruction, effectively predicting atomic conformations by integrating protein structure priors and handling unknown orientations.

Contribution

It develops a novel GNN-based autodecoder that incorporates geometric regularization and pose estimation, improving reconstruction accuracy over traditional MLP methods.

Findings

GNN outperforms MLP in synthetic datasets

Geometry-informed inductive bias enhances accuracy

Supports unknown orientations via ESL pose estimation

Abstract

We present a geometry-aware method for heterogeneous single-particle cryogenic electron microscopy (cryo-EM) reconstruction that predicts atomic backbone conformations. To incorporate protein-structure priors, we represent the backbone as a graph and use a graph neural network (GNN) autodecoder that maps per-image latent variables to 3D displacements of a template conformation. The objective combines a data-discrepancy term based on a differentiable cryo-EM forward model with geometric regularization, and it supports unknown orientations via ellipsoidal support lifting (ESL) pose estimation. On synthetic datasets derived from molecular dynamics trajectories, the proposed GNN achieves higher accuracy compared to a multilayer perceptron (MLP) of comparable size, highlighting the benefits of a geometry-informed inductive bias.