Hybrid Approach to Protein–Protein Complex Affinity Prediction Based on Language Models and Molecular Dynamics
International Journal of Molecular Sciences · 30.06.2026 · 27 (13) · 5925
DOI: 10.3390/ijms27135925 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Molecular Sciences, online erschienen am 30.06.2026. Verfasst von Elizaveta A Bogdanova und 2 weiteren Autor:innen. Beteiligte Einrichtung: AI Centre and Department of Biology, Lomonosov Moscow State University, Moscow 119991, Russia. Hauptthemen laut MEDLINE: Molecular Dynamics Simulation, Proteins. Die Arbeit nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: The Ministry of Economic Development of the Russian Federation.
Abstract (Originalsprache)
Protein–protein and protein–peptide interactions are fundamental to biological processes, making the accurate prediction of their binding affinity crucial for drug design and mutational analysis. Here, we develop HyBind-NN, a multimodal graph neural network that integrates protein language models (PLMs) with 3D structural and dynamic datasets to predict protein–protein and protein–peptide affinity. First, we demonstrate that combining ESM-2 sequence embeddings with precise 3D Voronoi spatial geometry enables accurate affinity predictions across diverse structural datasets. Next, we show that the inherent limitations of static rigid-body structures can be mitigated through a multi-task learning framework. By utilizing residue-level root mean square fluctuations (RMSF) derived from molecular dynamics (MD) as an auxiliary training target, the model implicitly learns to capture the conformational entropy of flexible peptides without requiring computationally expensive MD simulations during inference. In our benchmarking study, we observe that this multimodal architecture outperforms both purely sequence-based and strictly structural state-of-the-art algorithms, achieving a mean absolute error of 0.89 for pKD (1.12 kcal/mol for ∆G) on the independent benchmark. Finally, we confirmed through ablation analysis that while the PLM provides the dominant predictive signal, geometric representations and dynamic regularization are crucial for resolving subtle conformational rearrangements. This study highlights the synergistic potential of combining PLMs with physics-aware architectures and demonstrates their application towards the robust prediction of intermolecular binding affinity.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- AI Centre and Department of Biology, Lomonosov Moscow State University, Moscow 119991, Russia.
- Department of Information and Computer Technologies, Mendeleev University of Chemical Technology of Russia, Moscow 125047, Russia.
ORCID-Profile: Elizaveta A Bogdanova
Förderung
- The Ministry of Economic Development of the Russian Federation
Zitieren
APA
Bogdanova, E. A., Chernukhin, A. V., & Shaytan, A. K. (2026). Hybrid Approach to Protein–Protein Complex Affinity Prediction Based on Language Models and Molecular Dynamics. International Journal of Molecular Sciences, 27(13), 5925. https://doi.org/10.3390/ijms27135925
Vancouver
Bogdanova EA, Chernukhin AV, Shaytan AK. Hybrid Approach to Protein–Protein Complex Affinity Prediction Based on Language Models and Molecular Dynamics. International Journal of Molecular Sciences. 2026;27(13):5925. doi:10.3390/ijms27135925
BibTeX
@article{Bogdanova2026,
author = {Bogdanova, Elizaveta A and Chernukhin, Artem V and Shaytan, Alexey K},
title = {Hybrid Approach to Protein–Protein Complex Affinity Prediction Based on Language Models and Molecular Dynamics},
journal = {International Journal of Molecular Sciences},
year = {2026},
volume = {27},
number = {13},
pages = {5925},
doi = {10.3390/ijms27135925}
}
EUVIMED