Pool PaRTI: a PageRank-based pooling method for identifying critical residues and enhancing protein sequence representations
Bioinformatics · 2025
DOI: 10.1093/bioinformatics/btaf330 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bioinformatics, 2025. Verfasst von Alp Tartici und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Genetics, Stanford University, Palo Alto, CA 94304, United States. Hauptthemen laut MEDLINE: Proteins, Sequence Analysis, Protein, Software, Computational Biology. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NLM NIH HHS, NIH HHS, Chan Zuckerberg Biohub.
Abstract (Originalsprache)
Motivation Protein language models (PLMs) produce token-level embeddings for each residue, resulting in an output matrix with dimensions that vary based on sequence length. However, downstream machine learning models typically require fixed-length input vectors, necessitating a pooling method to compress the output matrix into a single vector representation of the entire protein. Traditional pooling methods often result in substantial information loss, impacting downstream task performance. We aim to develop a pooling method that produces more expressive general-purpose protein embedding vectors while offering biological interpretability. Results We introduce Pool PaRTI, a novel pooling method that leverages internal transformer attention matrices and PageRank to assign token importance weights. Our unsupervised and parameter-free approach consistently prioritizes residues experimentally annotated as critical for function, assigning them higher importance scores. Across four diverse protein machine learning tasks, Pool PaRTI enables significant performance gains in predictive performance. Additionally, it enhances interpretability by identifying biologically relevant regions without relying on explicit structural data or annotated training. To assess generalizability, we evaluated Pool PaRTI with two encoder-only PLMs, confirming its robustness across different models. Availability and implementation Pool PaRTI is implemented in Python with PyTorch and is available at github.com/Helix-Research-Lab/Pool_PaRTI.git. The Pool PaRTI sequence embeddings and residue importance values for all human proteins on UniProt are available at zenodo.org/records/15036725 for ESM2 and protBERT.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Department of Genetics, Stanford University, Palo Alto, CA 94304, United States.
- Department of Biomedical Data Science, Stanford University, Stanford, CA 94305, United States.
- Department of Bioengineering, Stanford University, Stanford, CA 94305, United States.
ORCID-Profile: Alp Tartici, Russ B Altman
Förderung
- NLM NIH HHS
- NIH HHS
- Chan Zuckerberg Biohub
- NIH NLM
- NIGMS NIH HHS
- Smith Fellowship
- NIH
Zitieren
APA
Tartici, A., Nayar, G., & Altman, R. B. (2025). Pool PaRTI: a PageRank-based pooling method for identifying critical residues and enhancing protein sequence representations. Bioinformatics. https://doi.org/10.1093/bioinformatics/btaf330
Vancouver
Tartici A, Nayar G, Altman RB. Pool PaRTI: a PageRank-based pooling method for identifying critical residues and enhancing protein sequence representations. Bioinformatics. 2025. doi:10.1093/bioinformatics/btaf330
BibTeX
@article{Tartici2025,
author = {Tartici, Alp and Nayar, Gowri and Altman, Russ B},
title = {Pool PaRTI: a PageRank-based pooling method for identifying critical residues and enhancing protein sequence representations},
journal = {Bioinformatics},
year = {2025},
doi = {10.1093/bioinformatics/btaf330}
}
EUVIMED