JanusDDG: a physics-informed neural network for sequence-based protein stability via two-fronts attention
Open Access
- 3 February 2026
- journal article
- research article
- Published by Springer Nature in Communications Biology
- Vol. 9 (1)
Abstract
Predicting how residue variations affect protein stability is crucial for rational protein design and for assessing the impact of disease-related mutations. Recent advances in protein language models have revolutionized computational protein analysis, enabling more accurate predictions of mutational effects. However, balancing predictive accuracy with the fundamental laws of thermodynamics remains a challenge for sequence-based models. Here we show JanusDDG, a physics-informed neural network that leverages embeddings from protein language models and a bidirectional cross-attention transformer architecture to predict stability changes for both single and multiple residue mutations. By adopting a physics-informed paradigm, the model is explicitly constrained to satisfy fundamental thermodynamic principles, such as antisymmetry and transitivity, while maintaining high predictive performance. Instead of conventional self-attention, JanusDDG employs a cross-interleaved attention mechanism that computes the relationship between wild-type and mutant embeddings to capture mutation-induced perturbations while preserving essential contextual information. Our results demonstrate that JanusDDG achieves state-of-the-art performance in predicting stability changes from sequence alone, matching or exceeding the accuracy of structure-based methods for both single and multiple mutations.Keywords
This publication has 47 references indexed in Scilit:
- Recent advances on protein engineering for improved stabilityBioDesign Research, 2025
- Protein stability models fail to capture epistatic interactions of double point mutationsProtein Science, 2024
- Deep Learning2024
- Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s NextJournal of Scientific Computing, 2022
- Stability Prediction for Mutations in the Cytosolic Domains of Cystic Fibrosis Transmembrane Conductance RegulatorJournal of Chemical Information and Modeling, 2021
- Computational Modeling of Protein Stability: Quantitative Analysis Reveals Solutions to Pervasive ProblemsStructure, 2020
- Enzymes for pharmaceutical and therapeutic applicationsBiotechnology and Applied Biochemistry, 2020
- Limitations and challenges in protein stability prediction upon genome variations: towards future applications in precision medicineComputational and Structural Biotechnology Journal, 2020
- Computational analysis of Alzheimer-causing mutations in amyloid precursor protein and presenilin 1Archives of Biochemistry and Biophysics, 2019
- VariBench: A Benchmark Database for VariationsHuman Mutation, 2012