INGA: protein function prediction combining interaction networks, domain assignments and sequence similarity

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Accurate prediction of protein function is essential for biological research and applications.
• Predicting Gene Ontology (GO) functional terms from protein sequence remains a significant computational challenge.
• Previous methods often struggle with the complexity and scale of functional annotation.

Purpose of the Study:

• To develop and present INGA, a novel web server for predicting protein function.
• To integrate multiple orthogonal approaches for improved functional term prediction accuracy.
• To provide a user-friendly and programmatically accessible tool for protein function prediction.

Main Methods:

• INGA combines sequence similarity searches and domain architecture analysis.
• Protein-protein interaction network data is integrated to refine predictions.
• Functional enrichment analysis is employed to derive consensus Gene Ontology (GO) term predictions.

Main Results:

• The INGA web server offers both a user-friendly web interface and programmatic RESTful services.
• Predictions are supported by evidence from annotating sequences, enhancing interpretability.
• INGA demonstrated consistent and strong performance on the CAFA-1 and CAFA-2 benchmark datasets.

Conclusions:

• INGA provides a robust and accurate method for predicting protein function using a multi-approach strategy.
• The tool is valuable for researchers needing to annotate protein functions computationally.
• The integration of diverse data sources leads to improved Gene Ontology (GO) term prediction accuracy.