PBSword

PBSword performs geometric comparison and search of protein-protein binding sites by representing each binding site as a high-dimensional vector of "visual words" that capture global and local shape features.

Key Features:

• Binding-site representation: Represents each protein-protein binding site as a high-dimensional vector composed of "visual words."
• Visual-word encoding: "Visual words" encapsulate both global and local shape features for nuanced characterization of binding sites.
• Scalable indexing: Employs a scalable indexing technique to enable rapid searches across the database.
• Database scope: Compares queries against a database of 194,322 domain-domain binding sites.
• Ranked results: Returns ranked matches based on similarity of visual word representations.
• Geometric similarity search: Performs precise comparisons and searches for geometrically similar protein-protein binding sites.

Scientific Applications:

• Binding-site similarity search: Identify geometrically similar protein-protein binding sites for structural comparison studies.
• Functional relationship exploration: Explore potential functional connections among proteins based on local structural features of binding sites.
• Protein-protein docking: Support analyses relevant to protein-protein docking by comparing binding-site geometries.
• Drug discovery: Enable detailed analyses of binding site geometries that inform drug discovery efforts.

Methodology:

Binding sites are encoded as high-dimensional vectors of "visual words" that capture global and local shape features, and a scalable indexing technique is used to search a database of 194,322 domain-domain binding sites and return ranked matches whose visual word representations closely match the query.