PCfun: a hybrid computational framework for systematic characterization of protein complex function

Area of Science:

• Molecular Biology
• Computational Biology
• Bioinformatics

Background:

• Protein complexes are crucial for biological functions, mediating cellular states and phenotypes.
• Identifying protein complex function is challenging despite advances in identifying their composition.

Purpose of the Study:

• To introduce PCfun, the first computational framework for systematic protein complex function annotation using Gene Ontology (GO) terms.
• To provide a novel tool for large-scale characterization of protein complex function.

Main Methods:

• Developed PCfun using word embeddings from 1 million PubMed Central articles and natural language processing.
• Employed an unsupervised nearest neighbor (NN) approach and a supervised Random Forest (RF) model.
• Integrated both approaches using a hypergeometric statistical test for GO term enrichment.

Main Results:

• PCfun accurately predicts protein complex functions by leveraging NLP and machine learning.
• The framework systematically annotates protein complex functions using GO terms.
• The study provides a computational framework to address the challenge of determining protein complex functions.

Conclusions:

• PCfun offers a novel paradigm for the large-scale characterization of protein complex function.
• The tool is expected to accelerate research in molecular biology and systems biology.
• PCfun is available as an open-source package for broader scientific use.