GaussianEM: Model compositional and conformational heterogeneity using 3D Gaussians

2025-12-25

AI Summary

To address the modeling challenge posed by the coexistence of continuous conformational motion and discrete states in cryo-EM data, this work introduces a pseudo-atomic representation framework based on 3D Gaussians, enabling the first unified modeling of compositional and conformational heterogeneity. Methodologically, we design a dual-encoder–single-decoder deep network architecture that differentiably parameterizes continuous conformational changes via Gaussian parameters, and jointly trains density reconstruction and structural refinement to establish an interpretable, consistent mapping between density and atomic models. Validation on both simulated and experimental cryo-EM data demonstrates that our method accurately reconstructs transition-state trajectories, preserves local atomic structural fidelity, enables atomic-resolution dynamic conformational analysis, and substantially advances the capability of flexible cryo-EM structure determination.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Understanding protein flexibility and its dynamic interactions with other molecules is essential for protein function study. Cryogenic electron microscopy (cryo-EM) provides an opportunity to directly observe macromolecular dynamics. However, analyzing datasets that contain both continuous motions and discrete states remains highly challenging. Here we present GaussianEM, a Gaussian pseudo-atomic framework that simultaneously models compositional and conformational heterogeneity from experimental cryo-EM images. GaussianEM employs a two-encoder-one-decoder architecture to map an image to its individual Gaussian components, and represent structural variability through changes in Gaussian parameters. This approach provides an intuitive and interpretable description of conformational changes, preserves local structural consistency along the transition trajectories, and naturally bridges the gap between density-based models and corresponding atomic models. We demonstrate the effectiveness of GaussianEM on both simulated and experimental datasets.

Problem

• Model compositional and conformational heterogeneity in cryo-EM data
• Analyze datasets with continuous motions and discrete states
• Bridge gap between density-based and atomic models

Innovation

• Uses 3D Gaussians to model heterogeneity
• Employs two-encoder-one-decoder architecture
• Bridges density-based and atomic models