iSPOT

iSPOT predicts peptide binding specificity across protein domain families by combining experimental interaction data and structural residue-residue contact information to score domain–peptide binding probabilities.

Key Features:

• Database construction: Extracts position-specific contacts from experimentally determined SH3/peptide and SH3/protein structures to build a multidimensional SH3-specific matrix of contact frequencies.
• Input requirements: Requires an SH3 domain sequence and a query decapeptide ligand to evaluate binding probabilities.
• Predictive capability: Computes the probability that a given peptide binds a specified SH3 domain and can score peptides across large sequence databases such as SWISS-PROT (approximately 5.6 x 10^9 decapeptides).
• Experimental validation: Validated by panning experiments that recovered natural interaction partners among top-ranked ligands in large peptide and protein sequence databases.
• Extensibility: Adapts the approach used for SH3 to other domain families including PDZ, WW, SH2, MHC, and EH for broader domain–ligand specificity prediction.
• Data foundations: Utilizes experimentally derived interaction data to build residue-residue contact frequency databases for PDZ, SH3, and WW domains.

Scientific Applications:

• Structural biology: Infers domain–ligand specificity and aids interpretation of structural determinants of binding.
• Drug discovery: Identifies and ranks candidate peptide ligands and potential therapeutic targets based on predicted binding probabilities.
• Molecular biology: Maps putative protein–protein interactions mediated by SH3 and other peptide-binding domains.
• Peptide therapeutics design: Supports selection and prioritization of peptide sequences for design of peptide-based drugs.

Methodology:

Extends the SPOT procedure by deriving position-specific residue–residue contact frequencies from known domain–ligand complexes, constructing a SH3-specific contact-frequency matrix, and computing peptide–domain binding probabilities to scan sequence databases.