The Bologna Annotation Resource (BAR 3.0): improving protein functional annotation.

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• Protein sequence analysis is crucial for understanding structure and function.
• Existing annotation resources require updates for increased data volume and query capabilities.

Purpose of the Study:

• To update the Bologna Annotation Resource (BAR) to version 3.0, enhancing its capabilities for predicting protein structural and functional features.
• To improve data volume, query functionalities, and user-provided information within the BAR resource.

Main Methods:

• Implemented a graph-based clustering of UniProtKB sequences using strict similarity criteria (≥40% identity, ≥90% coverage).
• Integrated annotations from UniProtKB, Gene Ontology (GO), Pfam, and Protein Data Bank (PDB) into sequence clusters.
• Developed an improved search interface supporting queries by UniProtKB accession, Fasta sequence, GO term, Pfam domain, organism, PDB, and ligand.

Main Results:

• BAR 3.0 contains over 28.8 million sequences organized into 1.36 million clusters.
• 22.2% of sequences have validated GO terms, and 47.4% have Pfam domains.
• 1.4% of clusters link to PDB structures and Hidden Markov Models for structural modeling, with significant performance improvement on CAFA2 targets.

Conclusions:

• BAR 3.0 provides a significantly enhanced resource for protein sequence annotation and feature prediction.
• The updated resource demonstrates state-of-the-art performance and offers advanced search capabilities for researchers.
• BAR 3.0 is publicly accessible, facilitating broader use in biological research and structural modeling.