Multiscale guidance of AlphaFold3 with heterogeneous cryo-EM data

2025-06-04

AI Summary

Current protein structure prediction models (e.g., AlphaFold3) excel at static, monomeric modeling but struggle to capture conformational heterogeneity in dynamic complexes. While cryo-EM can experimentally resolve native-state diversity, atomic model derivation from density maps remains reliant on heuristic, post-hoc refinement. To address this, we introduce CryoBoltz—a fine-tuning–free, AlphaFold3–based method for density-guided conformational sampling. During inference, it jointly enforces global and local density gradient constraints, employs multi-scale spatial mapping, and incorporates physics-based steric validation to enable end-to-end generation of multi-conformation atomic models directly from heterogeneous cryo-EM density maps. Evaluated on dynamic systems including transporters and antibodies, CryoBoltz significantly improves conformational coverage and atomic accuracy. It represents the first approach to achieve interpretable, experimentally verifiable modeling of functionally relevant conformations from high-resolution heterogeneous cryo-EM densities.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Protein structure prediction models are now capable of generating accurate 3D structural hypotheses from sequence alone. However, they routinely fail to capture the conformational diversity of dynamic biomolecular complexes, often requiring heuristic MSA subsampling approaches for generating alternative states. In parallel, cryo-electron microscopy (cryo-EM) has emerged as a powerful tool for imaging near-native structural heterogeneity, but is challenged by arduous pipelines to go from raw experimental data to atomic models. Here, we bridge the gap between these modalities, combining cryo-EM density maps with the rich sequence and biophysical priors learned by protein structure prediction models. Our method, CryoBoltz, guides the sampling trajectory of a pretrained protein structure prediction model using both global and local structural constraints derived from density maps, driving predictions towards conformational states consistent with the experimental data. We demonstrate that this flexible yet powerful inference-time approach allows us to build atomic models into heterogeneous cryo-EM maps across a variety of dynamic biomolecular systems including transporters and antibodies.

Problem

• Bridging cryo-EM data with protein structure prediction models
• Capturing conformational diversity in dynamic biomolecular complexes
• Building atomic models from heterogeneous cryo-EM maps

Innovation

• Combines cryo-EM data with protein structure prediction
• Uses global and local constraints from density maps
• Guides sampling trajectory of pretrained model