FISH

FISH identifies remote sequence homologues based on protein domains to determine family membership of domains in query protein sequences even at very low sequence identity to known homologues.

Key Features:

• Domain identification and annotation: Identifies domains within query protein sequences and assigns family membership for domain-level annotation.
• Remote homologue detection: Detects remote sequence homologues of protein domains with very low sequence identity to known homologues.
• Structural predictions: Produces probable two-dimensional (2D) and three-dimensional (3D) structure predictions for identified protein domains.
• Sequence alignments: Generates pairwise and multiple sequence alignments with homologues that have low sequence identity.
• Custom sequence searches and saHMM support: Accepts user-provided protein sequence collections and supports searches of public protein databases using specific saHMMs.
• Performance on SCOP: In a SCOP sequence test, achieved 99.3% accuracy for top hits aligning with the correct structural domain families (saHMMs) using an E-value cutoff of 0.1.

Scientific Applications:

• Domain annotation: Assigns domain family membership to support functional annotation of proteins.
• Structural modeling support: Informs 2D and 3D structural interpretation of protein domains, including those from remote homologues.
• Evolutionary and comparative analysis: Facilitates evolutionary studies and comparative analyses through alignments of distant homologues.
• Functional inference for low-identity proteins: Supports inference of functional properties for proteins with low sequence similarity to characterized homologues.

Methodology:

Searches employ structure-augmented HMMs (saHMMs) with an E-value cutoff of 0.1 and were evaluated on SCOP sequences (99.3% top-hit accuracy); the workflow generates pairwise and multiple sequence alignments and produces 2D and 3D structure predictions while allowing analyses of user-provided sequence collections and searches of public protein databases using saHMMs.