BAR-PLUS

BAR-PLUS performs functional and structural annotation of protein sequences by non-hierarchical clustering of large-scale pairwise sequence comparisons to enable reliable transfer of features within clusters.

Key Features:

• Large-Scale Sequence Comparison: Performs exhaustive pairwise sequence comparisons encompassing 13,495,736 protein chains and 988 complete proteomes.
• Non-hierarchical Clustering: Uses a non-hierarchical clustering procedure with a specialized metric that ensures reliable transfer of features within clusters.
• Annotation Data Sources: Derives annotations from UniProtKB, Gene Ontology (GO), Pfam, and the Protein Data Bank (PDB), with optional Structural Classification of Proteins (SCOP) information.
• Cluster-Based HMMs: Computes profile HMMs from sequence-to-structure alignments when PDB templates are present, producing a library of 10,858 Cluster-HMMs for aligning distantly related sequences.
• Pairwise Query Sequence-Structural Target Alignments: Provides pairwise query sequence–structural target alignments derived from corresponding Cluster-HMMs.
• Annotation Categories: Classifies clusters into categories: PDB with or without SCOP plus GO/Pfam; PDB without GO and/or Pfam; GO and/or Pfam without PDB; and no annotation.
• Statistical Significance of Annotations: Subdivides categories based on statistical significance of GO and Pfam annotations to indicate annotation reliability.

Scientific Applications:

• Structure Prediction: Predicting protein structures via sequence-to-structure alignments using Cluster-HMMs and PDB templates.
• Functional Annotation: Assigning functional annotations to proteins by transferring UniProtKB, GO, and Pfam terms within clusters.
• Evolutionary Analysis: Identifying evolutionary relationships by clustering and aligning distantly related sequences across complete proteomes.

Methodology:

Performs exhaustive pairwise sequence comparisons across 13,495,736 protein chains and 988 complete proteomes, applies a non-hierarchical clustering procedure with a specialized metric, computes profile HMMs from sequence-to-structure alignments when PDB templates are present to generate 10,858 Cluster-HMMs, and derives pairwise query sequence–structural target alignments from those Cluster-HMMs.