Patch-Surfer

Patch-Surfer predicts binding ligands for protein pockets by comparing local surface patch representations to a database of ligand-binding pockets to infer protein function and support drug discovery and off-target assessment.

Key Features:

• Local Surface Patch Analysis: Protein pockets are characterized as small local surface patches that capture physicochemical properties and enable identification of binding pockets despite lack of global shape similarity.
• 3D Zernike Descriptor (3DZD): Local patches are described using the 3D Zernike Descriptor to provide an efficient mathematical representation of patch features.
• Approximate Patch Position (APP): Approximate patch positions are encoded using a geodesic distance histogram to refine patch comparisons.
• Comprehensive Database: A large database of known ligand-binding pockets is used as a reference for querying and matching query pockets to ligands.

Scientific Applications:

• Protein function annotation: Predicting ligand binding to infer biological functions of proteins.
• Drug discovery and target identification: Identifying potential drug targets by matching query pockets to known ligand-binding sites.
• Off-target interaction and side-effect assessment: Predicting potential off-target ligand interactions to evaluate side-effect risks.
• Evolutionary and functional conservation studies: Recognizing similar binding pockets across different proteins to support studies of evolutionary relationships and functional conservation.

Methodology:

Protein pockets are represented as local surface patches described by 3D Zernike Descriptors (3DZD); approximate patch positions are encoded with geodesic distance histograms (APP); patch comparisons are performed against a database of ligand-binding pockets.