PatchSearch

PatchSearch identifies protein surface patches matching a query protein–ligand binding site to detect potential off-target interactions and propose plausible ligand binding modes with affinity estimates.

Key Features:

• Automated search workflow: Searches predefined or user-defined protein structure collections for binding sites compatible with the ligand's geometry and physicochemical properties.
• Non-sequential local alignment: Performs non-sequential local alignment to scan protein surfaces for patches matching the query binding site, relying on structural similarity rather than sequence homology.
• Ligand binding mode and affinity estimation: Proposes plausible ligand binding modes and estimates binding affinity using the Vinardo scoring function for identified off-target proteins.
• Controlled matching flexibility: Incorporates controlled flexibility via quasi-clique detection in product graphs that represent possible matchings between structures.
• Benchmarking and validation: Benchmarked on datasets ranging from 12 to over 7,000 protein structures across diverse ligands to evaluate early off-target identification performance.

Scientific Applications:

• Off-target identification: Identifies potential off-target interactions to aid mitigation of adverse drug effects.
• Drug repurposing: Supports drug repurposing by uncovering unexpected protein targets for existing drugs.
• Early-stage safety assessment: Predicts off-target interactions early in drug development to inform molecular modifications or dosage adjustments.
• Structural analysis: Contributes to structural understanding of protein–ligand interactions relevant to pharmacology and medicinal chemistry research.

Methodology:

Performs non-sequential local alignment to scan protein surfaces for matching patches; represents possible matchings as product graphs and detects quasi-cliques to allow controlled flexibility; proposes binding modes and scores affinities with the Vinardo scoring function; validated on datasets of 12 to over 7,000 protein structures.