SiteEngine

SiteEngine identifies protein binding-site regions by comparing local surface physico-chemical patterns without relying on sequence or fold similarities to predict molecular interactions and assist functional classification.

Key Features:

• Low-resolution surface representation: Represents protein surfaces by identifying chemically significant points to create a low-resolution model.
• Triangle hashing: Hashes triangles formed from chemically significant points using their physico-chemical properties.
• Non-sequence/fold-based similarity detection: Detects binding-site similarity independent of sequence or overall fold similarity.
• Global and local similarity identification: Enables efficient and rapid identification of both global and local similarities between binding sites across proteins.
• Large-Scale Search: Scans a database of complete protein structures to find regions similar to a given functional site.
• Comparative Analysis: Compares a potential binding site on a target protein with known binding sites to detect similarities.
• Discovery of Unknown Sites: Searches an entire protein structure for previously unknown functional sites that resemble known ones.

Scientific Applications:

• Prediction of molecular interactions: Infers likely binding interactions by matching surface physico-chemical patterns of binding sites.
• Functional classification of proteins: Assigns or refines protein function by identifying similarities to known functional sites.
• Identification of secondary drug-binding sites: Detects secondary binding sites on proteins that could explain off-target effects or side effects.
• Annotation of uncharacterized proteins: Predicts function for uncharacterized proteins by comparison to annotated binding sites.
• Drug design target discovery: Reveals novel functional sites on protein structures that resemble known ligand-binding sites for therapeutic exploration.

Methodology:

Identify chemically significant points on protein surfaces, construct a low-resolution surface model, hash triangles based on physico-chemical properties, and use those hashes to detect global and local similarities; apply these operations in database-wide scans, pairwise site comparisons, or whole-structure searches.