Protein Fold Classification at Scale: Benchmarking and Pretraining

2026-05-18

AI Summary

This work addresses the lack of large-scale, non-redundant benchmark datasets and limited model scalability in protein fold classification. To this end, the authors introduce TEDBench, the first large-scale, non-redundant benchmark constructed by clustering both AlphaFold-predicted and experimentally determined structures using Foldseek. They further propose Masked Invariant Autoencoders (MiAE), a self-supervised learning framework that integrates SE(3)-invariant graph neural networks with an aggressive masking strategy—up to 90% of residues masked—to efficiently learn structural representations. MiAE outperforms existing supervised methods and state-of-the-art baselines on TEDBench and demonstrates strong transferability on CATH v4.4.

Abstract

Classifying protein topology is essential for deciphering biological function, but progress is held back by the lack of large-scale benchmarks that avoid duplicates and by models that do not scale well. We introduce TEDBench, a large-scale, non-redundant benchmark for protein fold classification constructed from the Encyclopedia of Domains (TED) and Foldseek-clustered AlphaFold structures. We show that on TEDBench, current protein representation learning methods either require very large models or fail to deliver strong performance. To address this challenge, we propose Masked Invariant Autoencoders (MiAE), a self-supervised framework for protein structure representation learning. MiAE uses an extremely high masking ratio of up to 90% with an $\mathrm{SE(3)}$-invariant encoder and a lightweight decoder that reconstructs backbone coordinates from the latent representation and mask tokens. MiAE scales well and outperforms supervised counterparts and state-of-the-art baselines on TEDBench, establishing a strong recipe for protein fold classification. To test transfer beyond AlphaFold structures, we further benchmark on a curated dataset from experimental structures of CATH v4.4. TEDBench is available at https://github.com/BorgwardtLab/TEDBench.

Problem

• protein fold classification
• large-scale benchmark
• non-redundant dataset
• model scalability
• protein topology

Innovation

• protein fold classification
• self-supervised learning
• masked autoencoder
• SE(3)-invariance
• large-scale benchmark