Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks

2026-09-24

AI Summary

This study addresses the high computational cost of independently fitting implicit neural representations (INRs) to cryo-EM volumes and the resulting misalignment of representations across samples. To overcome these limitations, this work proposes Atelier, a self-supervised framework that pioneers the use of amortized INRs as geometry-aware analytical primitives. By leveraging a Transformer-based hypernetwork to amortize INR fitting, Atelier generates coordinate-conditioned continuous local feature fields that transcend the constraints of voxel grids and tokenizers, thereby facilitating 3D segmentation tasks. Pre-trained on 5,439 cryo-EM density maps, the proposed framework achieves high-fidelity reconstruction of multi-subunit assemblies and significantly outperforms existing baselines across eight voxel-level attribute prediction tasks.

Abstract

CryoEM map interpretation requires features that are spatially localized, consistent across samples, and informative across spatial scales. Most deep learning methods for map annotation extract features from fixed voxel grids. However, implicit neural representations (INRs) are able to model volumetric data as scale-agnostic, coordinate-conditioned functions. INRs are therefore attractive for cryoEM, but fitting a separate INR for each map is too expensive for large-scale feature extraction and produces representations that are not aligned across samples. We introduce Atelier, a self-supervised framework that amortizes INR fitting for reconstructed cryoEM maps. Pretrained on 5,439 Electron Microscopy Data Bank maps, Atelier is a transformer-based hypernetwork that generates high-fidelity reconstructions across a wide range of protein structures, including large multi-subunit assemblies. Beyond reconstruction, the INR generated by the pretrained transformer exposes a continuous, local feature field through its intermediate activations at any spatial query point, a property that voxel grid and patch-tokenizer architectures do not naturally provide. Used as auxiliary channels to a 3D nested U-Net annotation head trained from scratch, these coordinate-conditioned features improve performance on eight voxel-level property prediction tasks over a volume-only baseline. Our results demonstrate that amortized implicit neural representations are an effective primitive for geometry-aware analysis of cryoEM data.

Problem

• CryoEM
• Implicit Neural Representations
• Self-Supervised Learning
• Feature Extraction
• Hypernetworks

Innovation

• Implicit Neural Representations
• Hypernetworks
• Self-Supervised Learning
• CryoEM
• Amortized Inference