AutoSite

AutoSite identifies and characterizes ligand-binding sites on proteins with known three-dimensional structures to support computational drug design, lead optimization, and protein function assignment.

Key Features:

• Identification of Binding Sites: Identifies ligand-binding sites as clusters of 3D points termed "fills", with each point labeled as hydrophobic or as a hydrogen-bond donor or acceptor.
• Feature Point Derivation: Derives feature points from fills that represent putative positions for ligand atoms involved in hydrophobic interactions and hydrogen bonding.
• Energetic Selection and Clustering: Uses an energetic approach to select high-affinity points around a receptor and applies clustering techniques to segregate these points into fills.
• Comparative Accuracy: Demonstrates higher accuracy than other leading methods and produces fills that more closely match the shape and properties of actual ligands compared to AutoLigand.
• Performance Metrics: On the Astex Diverse Set, identified 79% of hydrophobic ligand atoms, predicted 81% of hydrogen acceptor and 62% of hydrogen donor ligand atoms interacting with receptors, and predicted 81.2% of water molecules that mediate ligand–receptor interactions.

Scientific Applications:

• Structure-based drug design: Provides predicted binding pockets and feature points for modeling ligand–receptor interactions in structure-based drug design.
• Lead optimization: Supplies feature points that guide optimization of small-molecule leads to improve binding specificity and efficacy.
• Protein function assignment: Helps assign protein function by identifying potential ligand-binding sites on 3D protein structures.

Methodology:

AutoSite employs an energetic approach to select high-affinity points around a receptor's structure, utilizing clustering techniques to segregate these points into fills.