PSI-Search API (EBI)

PSI-Search API (EBI) performs protein similarity searches by combining SSEARCH (Smith–Waterman) local alignments with PSI-BLAST profile construction to detect homologs with improved selectivity and maintained sensitivity.

Key Features:

• Optimal Smith–Waterman Local Alignment: Uses SSEARCH to perform optimal Smith–Waterman local alignments for precise sequence similarity detection.
• PSI-BLAST Profile Construction: Builds position-specific scoring matrices (PSSMs) using the PSI-BLAST profile construction strategy to enhance homolog detection.
• Homologous Over-Extension (HOE) Mitigation: Implements an optional sequence boundary-masking procedure to prevent profile contamination by alignments that extend into unrelated domains.
• Enhanced Selectivity and Sensitivity: Achieves approximately four-fold better selectivity while maintaining similar sensitivity at 50% and 60% family coverage compared with traditional PSI-BLAST.
• Reduced False Positives: Produces 2- to 4-fold fewer false positives in comparisons with JackHMMER.

Scientific Applications:

• Evolutionary Biology: Enables detection of distant homologs to inform evolutionary relationships.
• Protein Functional Annotation: Supports identification of homologous sequences for transfer or inference of functional annotation.
• Conserved Domain Identification: Assists in detecting conserved protein domains and mitigating domain boundary over-extension.

Methodology:

Performs optimal Smith–Waterman alignments with SSEARCH, constructs PSSMs using the PSI-BLAST profile construction strategy, applies optional sequence boundary-masking to prevent homologous over-extension (HOE), and compares selectivity/sensitivity metrics to PSI-BLAST and false-positive rates to JackHMMER.