DoGSiteScorer
DoGSiteScorer is a grid-based tool for the automated detection, characterization, and druggability assessment of protein binding pockets. It applies a Difference of Gaussian filter to the three-dimensional protein structure to identify potential pockets and subdivide them into sub-pockets. For each predicted pocket, the tool calculates descriptors covering size, shape, enclosure, and chemical properties.
DoGSiteScorer provides two complementary druggability assessments: a simple score based on pocket volume, hydrophobicity, and enclosure, and a support vector machine model using a broader set of pocket descriptors. Scores range from zero to one, with higher values indicating a greater estimated likelihood that the pocket can bind drug-like molecules.
Key benefits
Fully automated detection of binding pockets and sub-pockets
Requires only the three-dimensional protein structure
Calculates geometric and physicochemical pocket descriptors
Provides interpretable druggability scores between zero and one
Combines a simple descriptor-based score with machine-learning prediction
Supports rapid comparison and prioritization of potential binding sites
Applications
Identification of potential small-molecule binding sites
Druggability assessment of protein pockets
Selection of binding sites for docking and virtual screening
Support for target assessment in early-stage drug discovery
Structural characterization of protein cavities
Intended use
DoGSiteScorer is intended for structural biologists, medicinal chemists, computational chemists, and researchers in structure-based drug discovery who need to identify and prioritize potential ligand-binding pockets. It is particularly suited for users who want an automated, structure-based assessment of pocket geometry, physicochemical properties, and estimated druggability.
Contact:
Website