BiteNet

BiteNet detects protein binding sites at large scale by treating protein conformational ensembles as dynamic 3D-images (3D-videos) to enable spatiotemporal analysis of binding-site formation and allostery.

Key Features:

• Large-Scale Detection: Performs high-throughput identification of protein binding sites across extensive conformational datasets.
• Conformation-Based Analysis: Represents three-dimensional protein conformations as 3D-images, aligning binding-site identification with object-detection concepts from computer vision.
• Spatiotemporal Detection: Analyzes temporal and spatial information in conformational ensembles to detect conformation-specific and allosteric binding sites in proteins such as the epidermal growth factor receptor, ion channels, and G protein-coupled receptors.
• Performance Superiority: Processes ~1000 conformations of a ~2000-atom protein in about 1.5 minutes, offering improved accuracy and speed compared to existing methods.

Scientific Applications:

• Druggable Genome Expansion: Identifies novel binding sites to expand the druggable genome.
• Allosteric Drug Discovery: Enables discovery and targeting of allosteric sites implicated in protein regulatory functions for drug discovery.
• Therapeutic Design Support: Provides comprehensive information on protein dynamics and interaction sites to inform design of therapeutic agents.

Methodology:

BiteNet treats protein conformations as 3D-images and binding sites as detectable objects, processing conformational ensembles as 3D-videos to perform spatiotemporal examination of binding-site appearance.