CATH FunFHMMer web server: protein functional annotations using functional family assignments.

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• The increasing volume and diversity of sequenced proteins create a significant challenge for accurate protein function annotation.
• Multidomain proteins, with their combinatorial domain arrangements, further complicate the understanding of protein sequence-structure-function relationships.

Purpose of the Study:

• To introduce the FunFHMMer web server for predicting Gene Ontology (GO) annotations of protein sequences.
• To leverage the CATH-Gene3D resource for domain-based functional classification and prediction of functional sites.

Main Methods:

• Utilized domain-based functional classification from the CATH-Gene3D resource.
• Developed the FunFHMMer web server for querying protein sequences and generating GO annotations.
• Validated predictive accuracy against established methods like BLAST, Pfam, and CDD.

Main Results:

• FunFHMMer demonstrated superior performance in predicting GO annotations compared to BLAST, Pfam, and CDD on a test set of 95 proteins.
• Independent validation ranked FunFHMMer among the top methods for predicting both Biological Process and Molecular Function Ontologies.
• The server provides valuable information for predicting functional sites.

Conclusions:

• FunFHMMer is a highly effective tool for protein function prediction, particularly for addressing the challenge of annotation gaps.
• The web server offers a valuable resource for researchers needing accurate GO annotations and functional site predictions.
• Its performance validates the domain-based approach for understanding protein function.