SPACE: STRING proteins as complementary embeddings
Bioinformatics · 01.09.2025 · 41 (9)
DOI: 10.1093/bioinformatics/btaf496 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bioinformatics, online erschienen am 09.09.2025. Verfasst von Dewei Hu und 3 weiteren Autor:innen. Beteiligte Einrichtung: Novo Nordisk Foundation Center for Protein Research, Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, Universit…. Hauptthemen laut MEDLINE: Proteins, Computational Biology. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Novo Nordisk Foundation, Swiss Institute of Bioinformatics.
Abstract (Originalsprache)
Motivation Representation learning has revolutionized sequence-based prediction of protein function and subcellular localization. Protein networks are an important source of information complementary to sequences, but the use of protein networks has proven to be challenging in the context of machine learning, especially in a cross-species setting. Results We leveraged the STRING database of protein networks and orthology relations for 1322 eukaryotes to generate network-based cross-species protein embeddings. We did this by first creating species-specific network embeddings and subsequently aligning them based on orthology relations to facilitate direct cross-species comparisons. We show that these aligned network embeddings ensure consistency across species without sacrificing quality compared to species-specific network embeddings. We also show that the aligned network embeddings are complementary to sequence embedding techniques, despite the use of sequence-based orthology relations in the alignment process. Finally, we validated the embeddings by using them for two well-established tasks: subcellular localization prediction and protein function prediction. Training logistic regression classifiers on aligned network embeddings and sequence embeddings improved the accuracy over using sequence alone, reaching performance numbers close to state-of-the-art deep-learning methods. Availability and implementation The source code and scripts for generating the network-based cross-species protein embeddings are available at https://github.com/deweihu96/SPACE. Precomputed network embeddings and sequence embeddings for all eukaryotic proteins are included in STRING version 12.0 (https://string-db.org/cgi/download).
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Novo Nordisk Foundation Center for Protein Research, Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen 2200, Denmark.
- Department of Molecular Life Sciences, University of Zurich, Zurich 8057, Switzerland.
- SIB Swiss Institute of Bioinformatics, Amphipôle, Quartier UNIL-Sorge, Lausanne 1015, Switzerland.
- ZS Discovery, ZS Associates, Kongens Lyngby 2800, Denmark.
ORCID-Profile: Dewei Hu, Christian von Mering
Förderung
- Novo Nordisk Foundation
- Swiss Institute of Bioinformatics
Zitieren
APA
Hu, D., Szklarczyk, D., von Mering, C., & Jensen, L. J. (2025). SPACE: STRING proteins as complementary embeddings. Bioinformatics, 41(9). https://doi.org/10.1093/bioinformatics/btaf496
Vancouver
Hu D, Szklarczyk D, von Mering C, Jensen LJ. SPACE: STRING proteins as complementary embeddings. Bioinformatics. 2025;41(9). doi:10.1093/bioinformatics/btaf496
BibTeX
@article{Hu2025,
author = {Hu, Dewei and Szklarczyk, Damian and von Mering, Christian and Jensen, Lars Juhl},
title = {SPACE: STRING proteins as complementary embeddings},
journal = {Bioinformatics},
year = {2025},
volume = {41},
number = {9},
doi = {10.1093/bioinformatics/btaf496}
}
EUVIMED