StabilityArc: Decoding Protein Sequence Embeddings into Generalizable Stability Landscapes
Abstract
Every protein has a unique stability landscape, but the physical consequences of mutation are governed by recurring biochemical constraints. We test whether a shared decoder, trained on measurements from diverse proteins, can interpret these constraints in an unseen target, enabling cross-protein transfer for initial experimental round prescreening. We present StabilityArc , which maps frozen ESMC-600M residue representations through a shared RoPE transformer to an Lx20 matrix of substitution effects; a symmetric, contact-aware residual aids in predicting epistasis in simultaneous substitutions. In 66 strict leave-one-protein-out evaluations covering 134,794 ProteinGym variants, StabilityArc achieves 0.7134 Spearman correlation, exceeding the strongest zero-shot baseline, ProSST-2048 (0.6526), by 0.0608. We further explore the utility of this method by providing the score as a prior for Kermut, achieving Spearman correlation of 0.8280 across three supervised split schemes, improving on Kermut's reported 0.8167.
Related
- Decoding Stability and Selectivity in Unnatural DNA Base Pairs with Spatially Separated Recognition Interfaces.
- An open structure-guided RNA foundation model for generalizable prediction
- A generalizable Hi-C foundation model for chromatin architecture, single-cell and multiomics analysis across species
- Dynamic graph prior-guided latent adaptation for generalizable single-cell perturbation response prediction.
- A generalizable method for the purification and biophysical characterization of bacterial membrane receptors.
BibTeX
@misc{feller2026stabilityarc,
title = {{StabilityArc: Decoding Protein Sequence Embeddings into Generalizable Stability Landscapes}},
author = {Aaron L. Feller and Andrew D. Ellington and Claus O. Wilke},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.00742},
eprint = {2610.00742},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.00742},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - StabilityArc: Decoding Protein Sequence Embeddings into Generalizable Stability Landscapes
AU - Aaron L. Feller
AU - Andrew D. Ellington
AU - Claus O. Wilke
PY - 2026
DA - 2026/10/15
PB - arXiv
DO - 10.48550/arxiv.2610.00742
UR - https://arxiv.org/abs/2610.00742
AB - Every protein has a unique stability landscape, but the physical consequences of mutation are governed by recurring biochemical constraints. We test whether a shared decoder, trained on measurements from diverse proteins, can interpret these constraints in an unseen target, enabling cross-protein transfer for initial experimental round prescreening. We present StabilityArc , which maps frozen ESMC-600M residue representations through a shared RoPE transformer to an Lx20 matrix of substitution effects; a symmetric, contact-aware residual aids in predicting epistasis in simultaneous substitutions. In 66 strict leave-one-protein-out evaluations covering 134,794 ProteinGym variants, StabilityArc achieves 0.7134 Spearman correlation, exceeding the strongest zero-shot baseline, ProSST-2048 (0.6526), by 0.0608. We further explore the utility of this method by providing the score as a prior for Kermut, achieving Spearman correlation of 0.8280 across three supervised split schemes, improving on Kermut's reported 0.8167.
ER -