MutaStruct: Revisiting Benchmarks for Single-Site Mutation-Induced Structural Change Prediction
- 1The University of Alabama at Birmingham, Alabama, United States
Abstract
Predicting the structural consequences of single-site mutations is fundamental to understanding protein function, disease mechanisms, and protein engineering. However, existing benchmarks for mutation-induced structural change prediction are often affected by duplicate mutation events, highly similar structural pairs, and experimental-state confounders, which can inflate reported performance. We introduce MutaStruct, a curated benchmark of experimentally observed single-mutation protein structure pairs designed for rigorous evaluation of mutation-induced structural changes. MutaStruct combines residue-level sequence validation, coordinate-level quality control, context-aware filtering, event-level deduplication, and leakage-aware data splitting to improve dataset reliability and evaluation rigor. We benchmark sequence-only structure prediction, physics-based refinement, and learning-based refinement methods together with a Copy-WT diagnostic control. Copy-WT remains highly competitive on backbone-based metrics, while method rankings change substantially across local and mutation-site measures. These results show that overall structural fidelity does not necessarily imply accurate modeling of mutation-associated structural change, underscoring the need for further methodological development beyond current paradigms. The benchmark dataset is currently undergoing further curation and expansion and will be released in a future journal-version update. The codebase, trained models, and evaluation pipeline are publicly available at https://github.com/HySonLab/MutaStruct
Related
- RPFace: From biased structural benchmarks to generalizable RNA-protein interface prediction
- Evaluating Autonomous LLM Agents Across Molecular Prediction and Optimization Benchmarks
- Measurement reliability bounds functional benchmarks and relocates where variant effect prediction fails
- MutCleaner: Cleaning and Standardizing Biological Mutation Datasets for Variant Effect Prediction
- Inherent limitations of zero-shot mutation prediction in protein engineering
BibTeX
@misc{huynh2026mutastruct,
title = {{MutaStruct: Revisiting Benchmarks for Single-Site Mutation-Induced Structural Change Prediction}},
author = {T.-P. Huynh and D.-M. Ho and V. T. D. Nguyen and Q.-H. Kha and N.-T. Pham and K.-H. Lai and T.-S. Hy},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.21.760837},
url = {https://www.biorxiv.org/content/10.64898/2026.10.21.760837v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - MutaStruct: Revisiting Benchmarks for Single-Site Mutation-Induced Structural Change Prediction
AU - T.-P. Huynh
AU - D.-M. Ho
AU - V. T. D. Nguyen
AU - Q.-H. Kha
AU - N.-T. Pham
AU - K.-H. Lai
AU - T.-S. Hy
PY - 2026
DA - 2026/10/21
PB - bioRxiv
DO - 10.64898/2026.10.21.760837
UR - https://www.biorxiv.org/content/10.64898/2026.10.21.760837v1
AB - Predicting the structural consequences of single-site mutations is fundamental to understanding protein function, disease mechanisms, and protein engineering. However, existing benchmarks for mutation-induced structural change prediction are often affected by duplicate mutation events, highly similar structural pairs, and experimental-state confounders, which can inflate reported performance. We introduce MutaStruct, a curated benchmark of experimentally observed single-mutation protein structure pairs designed for rigorous evaluation of mutation-induced structural changes. MutaStruct combines residue-level sequence validation, coordinate-level quality control, context-aware filtering, event-level deduplication, and leakage-aware data splitting to improve dataset reliability and evaluation rigor. We benchmark sequence-only structure prediction, physics-based refinement, and learning-based refinement methods together with a Copy-WT diagnostic control. Copy-WT remains highly competitive on backbone-based metrics, while method rankings change substantially across local and mutation-site measures. These results show that overall structural fidelity does not necessarily imply accurate modeling of mutation-associated structural change, underscoring the need for further methodological development beyond current paradigms. The benchmark dataset is currently undergoing further curation and expansion and will be released in a future journal-version update. The codebase, trained models, and evaluation pipeline are publicly available at https://github.com/HySonLab/MutaStruct
ER -