PHMMER

PHMMER performs sensitive homology searches of protein sequences against protein sequence databases using profile Hidden Markov Models (HMMs) to detect close and remote homologs for functional and evolutionary inference.

Key Features:

• Profile Hidden Markov Models (HMMs): Uses profile HMMs that capture sequence variability within protein families to increase sensitivity for homolog detection.
• Iterative Search Algorithm: Incorporates the jackhmmer iterative search to refine profiles across multiple iterations and identify distant evolutionary relationships.
• Expanded Libraries: Leverages profile HMM libraries derived from CATH-Gene3D, PIRSF, Superfamily, and TIGRFAMs for domain and family annotation.
• Result Presentation: Produces taxonomic distribution and domain architecture summaries to summarize homology search results.
• Taxonomic Filtering: Supports restricting searches to specified taxonomic groups to focus results and improve specificity.
• Benchmarked Performance: Demonstrates superior accuracy for remote homolog detection compared with profile-based tools (CS-BLAST, HHSEARCH) and non-profile tools (NCBI-BLAST, USEARCH) while maintaining reasonable computational cost.

Scientific Applications:

• Homology Detection: Identify homologous protein sequences across diverse databases to support annotation and evolutionary studies.
• Functional Inference: Aid prediction of functions for uncharacterized proteins by detecting remote homologs and transferring annotations from related sequences.
• Domain Architecture Analysis: Support analysis of protein domain architectures via HMM matches and domain summaries.

Methodology:

PHMMER employs profile HMMs constructed from multiple sequence alignments of protein families, searches these models against target protein databases to identify matching sequences, and uses the jackhmmer iterative algorithm to update profiles with newly identified hits across iterations.