BAR

BAR annotates protein sequences with structural and functional information by clustering UniProtKB sequences and transferring statistically validated annotations including Gene Ontology (GO), Pfam domains, and PDB-derived structural models.

Key Features:

• Graph-Based Clustering: Implements a graph-based, non-hierarchical clustering that groups UniProtKB sequences using pairwise similarity thresholds of sequence identity ≥40% and alignment coverage ≥90%.
• Comprehensive Annotation Database: Aggregates annotations from UniProtKB, Gene Ontology (GO), Pfam domains, and Protein Data Bank (PDB) structures within clusters.
• Large-Scale Sequence Analysis: Processes 28,869,663 sequences organized into 1,361,773 clusters; 22.2% of sequences have at least one validated GO term, 47.4% have at least one Pfam domain, and 1.4% of clusters include PDB structures.
• Cluster-HMM for Structural Modeling: Associates a hidden Markov model (Cluster-HMM) to clusters containing PDB structures to enable template-target alignments and direct computation of 3D models from sequence data.
• Statistical Validation of Annotations: Applies statistical validation to annotation transfer within clusters to ensure reliable assignment of GO terms and Pfam domains even at low homology.

Scientific Applications:

• Functional Annotation: Assigns validated Gene Ontology (GO) terms and Pfam annotations to infer protein biological roles.
• Structural Modeling: Uses Cluster-HMMs and PDB-derived templates to compute 3D structural models for cluster members.
• Comparative Genomics: Enables large-scale cross-comparison of UniProtKB sequences to investigate evolutionary relationships among proteins.

Methodology:

Performs graph-based, non-hierarchical clustering of UniProtKB sequences using pairwise thresholds of ≥40% sequence identity and ≥90% alignment coverage; computes profile Hidden Markov Models from sequence-to-structure alignments; associates Cluster-HMMs to clusters with PDB structures for template-target alignment and 3D model computation; and applies statistical validation for annotation transfer.