Abstract:Recent advances in generative modeling show that pretrained representations can improve generation as conditioning features or alignment targets. Motivated by this, we study protein representations for predicting structures beyond conventional function annotation. We propose TriProRep, a structure-aware pretraining method that jointly models three aligned residue-level views: amino-acid identity, backbone geometry, and local full-atom geometry, discretely encoded via VQ-VAE tokenizers. By pretraining to recover original tokens from generator-corrupted views, TriProRep learns to distinguish plausible but incorrect cross-view augmentations from the original protein. We further introduce RepSP, a benchmark for evaluating protein representations in structure-predictive settings. RepSP tests three uses of representations: homodimer co-folding from apo-chain representations, residue-level prediction of homodimer-derived interaction properties, and representation-aligned monomer structure prediction. Across these tasks, TriProRep improves over sequence-only and prior structure-aware representation models, while maintaining competitive performance on conventional benchmarks.
| Subjects: | Biomolecules (q-bio.BM); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.22133 [q-bio.BM] |
| (or arXiv:2605.22133v1 [q-bio.BM] for this version) | |
| https://doi.org/10.48550/arXiv.2605.22133 arXiv-issued DOI via DataCite (pending registration) |
From: Taewon Kim [view email]
[v1]
Thu, 21 May 2026 08:07:36 UTC (5,215 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。