An experimental analysis of graph representation learning for Gene Ontology based protein function prediction

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate protein function prediction is vital for understanding biological systems and biomedical applications.
• Computational methods, particularly graph representation learning, are increasingly used to predict protein functions from sequence data.
• Existing methods face challenges in handling the vast and growing amount of protein sequence information.

Purpose of the Study:

• To review fundamental concepts in graph embedding algorithms for protein function prediction.
• To describe graph representation learning methods based on PPI networks, protein structure, Gene Ontology graphs, and integrated data.
• To summarize and analyze commonly used approaches, their results, limitations, and future directions.

Main Methods:

• Review of graph embedding algorithms and their application to protein function prediction.
• Categorization of methods based on data sources: PPI network, protein structure, Gene Ontology graph, and integrated graph.
• Detailed explanation and diagramming of commonly used approaches for each data category.

Main Results:

• Summary of various graph representation learning techniques applied to different biological data types.
• Detailed explanation of the performance and outcomes of different methods.
• Identification of limitations and potential solutions for current protein function prediction models.

Conclusions:

• Graph representation learning offers a powerful framework for advancing protein function prediction.
• Integration of diverse data sources, including PPI networks and protein structures, improves prediction accuracy.
• Future research should focus on addressing current limitations and exploring novel graph embedding strategies.