PfamScan API

PfamScan API annotates catalytic and active-site residues in Pfam protein families by using Pfam Hidden Markov Models (HMMs) to search FASTA sequences and transfer experimentally determined catalytic residue annotations across aligned family members.

Key Features:

• Pfam HMM FASTA search: Uses Pfam library Hidden Markov Models to search FASTA sequences as the basis for annotation transfer.
• Active Site Annotation: Predicts and annotates catalytic residues across Pfam families to increase coverage of experimentally characterized active sites.
• Rule-Based Prediction Methodology: Applies a strict set of rules to transfer experimental catalytic-site annotations between aligned family members while minimizing false positives.
• Extensive Database Creation: Generates a database containing 606,110 predicted active-site residues, with 94% of these not previously documented in UniProtKB.
• High Sensitivity and Specificity: Methodology was assessed for sensitivity and specificity, estimating approximately 3% false positives among predictions.
• Comparative Analysis: Compares predictions with UniProtKB, Catalytic Site Atlas, PROSITE, and MEROPS to identify novel predictions and limitations in other resources.
• Scalability and Flexibility: Applies the data-transfer methodology to any alignment that has associated experimental active-site information.
• Regular Updates: Recalculates active-site predictions at each Pfam release to maintain currency of the annotation set.

Scientific Applications:

• Enzyme Function Prediction: Identifies catalytic residues in uncharacterized proteins within enzymatic Pfam families to support inference of enzymatic functions.
• Protein Engineering and Drug Design: Provides residue-level active-site annotations to inform protein engineering strategies and the design of inhibitors or activators targeting specific catalytic residues.
• Comparative Genomics: Enables analysis of active-site conservation across species and the study of evolutionary relationships among proteins.

Methodology:

Searches FASTA sequences with Pfam HMMs and transfers experimentally determined catalytic residue annotations across aligned family members using a strict rule set; predictions were compiled into a database, assessed for sensitivity and specificity (≈3% estimated false positives), and are recalculated at each Pfam release.