Deep learning models for unbiased sequence-based PPI prediction plateau at an accuracy of 0.65
Bioinformatics · 01.07.2025 · 41 (Supplement_1) · i590-i598
DOI: 10.1093/bioinformatics/btaf192 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bioinformatics, online erschienen am 15.07.2025. Verfasst von Timo Reim und 4 weiteren Autor:innen. Beteiligte Einrichtung: Data Science in Systems Biology, TUM School of Life Sciences, Technical University of Munich, Freising, 85354, Germany. Hauptthemen laut MEDLINE: Deep Learning, Protein Interaction Mapping, Computational Biology, Proteins. Die Arbeit wurde bisher 17-mal zitiert (Crossref) und nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: German Federal Ministry of Education and Research, Klaus Tschira Stiftung, Deutsche Forschungsgemeinschaft.
Abstract (Originalsprache)
Motivation As most proteins interact with other proteins to perform their respective functions, methods to computationally predict these interactions have been developed. However, flawed evaluation schemes and data leakage in test sets have obscured the fact that sequence-based protein–protein interaction (PPI) prediction is still an open problem. Recently, methods achieving better-than-random performance on leakage-reduced PPI data have been proposed. Results Here, we show that the use of ESM-2 protein embeddings explains this performance gain irrespective of model architecture. We compared the performance of models with varying complexity, per-protein, and per-token embeddings, as well as the influence of self- or cross-attention, where all models plateaued at an accuracy of 0.65. Moreover, we show that the tested sequence-based models cannot implicitly learn a contact map as an intermediate layer. These results imply that other input types, such as structure, might be necessary for producing reliable PPI predictions. Availability and implementation All code for models and execution of the models is available at https://github.com/daisybio/PPI_prediction_study. Python version 3.8.18 and PyTorch version 2.1.1 were used for this study. The environment containing the versions of all other packages used can be found in the GitHub repository. The used data are available at https://doi.org/10.6084/m9.figshare.21591618.v3.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Data Science in Systems Biology, TUM School of Life Sciences, Technical University of Munich, Freising, 85354, Germany.
- Department Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, 91052, Germany.
- Munich Data Science Institute (MDSI), Technical University of Munich, Garching bei München, 85748, Germany.
ORCID-Profile: Timo Reim, Anne Hartebrodt, David B Blumenthal, Judith Bernett, Markus List
Förderung
- German Federal Ministry of Education and Research
- Klaus Tschira Stiftung
- Deutsche Forschungsgemeinschaft
- German Research Foundation
Zitieren
APA
Reim, T., Hartebrodt, A., Blumenthal, D. B., Bernett, J., & List, M. (2025). Deep learning models for unbiased sequence-based PPI prediction plateau at an accuracy of 0.65. Bioinformatics, 41(Supplement_1), i590-i598. https://doi.org/10.1093/bioinformatics/btaf192
Vancouver
Reim T, Hartebrodt A, Blumenthal DB, Bernett J, List M. Deep learning models for unbiased sequence-based PPI prediction plateau at an accuracy of 0.65. Bioinformatics. 2025;41(Supplement_1):i590-i598. doi:10.1093/bioinformatics/btaf192
BibTeX
@article{Reim2025,
author = {Reim, Timo and Hartebrodt, Anne and Blumenthal, David B and Bernett, Judith and List, Markus},
title = {Deep learning models for unbiased sequence-based PPI prediction plateau at an accuracy of 0.65},
journal = {Bioinformatics},
year = {2025},
volume = {41},
number = {Supplement_1},
pages = {i590--i598},
doi = {10.1093/bioinformatics/btaf192}
}
EUVIMED