Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks
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.
Related
- Online Supervised Dimension Reduction with Random Features: Diagnostics and Computational Trade-offs
- Semi-supervised learning on singular data sets
- Corrections to ``The Method of Gaps: Exact Expressions for the Generalization Error of Supervised Learning Algorithms
- Neural Supervised Clustering for Learning Individualized Treatment Rules with Many Treatments
- Rethinking Representation Geometry in Supervised Contrastive Learning under Long-Tailed Data
BibTeX
@misc{lo2026atelier,
title = {{Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks}},
author = {Phillip Lo and Sudarshan Babu and Dari Kimanius and Aly A. Khan},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.30569},
eprint = {2609.30569},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.30569},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks
AU - Phillip Lo
AU - Sudarshan Babu
AU - Dari Kimanius
AU - Aly A. Khan
PY - 2026
DA - 2026/09/24
PB - arXiv
DO - 10.48550/arxiv.2609.30569
UR - https://arxiv.org/abs/2609.30569
AB - 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.
ER -