DoGSite3

DoGSite3 predicts robust and reliable small molecule binding sites and computes their geometrical and chemical descriptors for early phase drug discovery involving experimental or predicted protein structure models.

Key Features:

• Grid-based binding site detection: DoGSite3 uses a Difference-of-Gaussian filter to detect cavities on the protein surface.
• Sub-pocket prediction: The tool predicts pockets and their sub-pockets based on the DoGSite algorithm.
• Rotation and translation invariance: A normalization procedure before binding site prediction makes the tool largely rotation- and translation-invariant.
• Ligand-biased grid calculation: Known ligands in the structure can be used to bias the grid by sufficiently buried ligand fragments.
• Novel chemical descriptors: The output encompasses chemical binding site descriptors considering solvent accessibility.
• Optimized parameters: Comprehensive parameter optimization provides increased robustness and reliability compared to its predecessor.
• Fast runtime: DoGSite3 runs finish within seconds.

Scientific Applications:

• Early phase drug discovery: Binding site prediction on protein structures supports drug discovery when experimental or predicted structure models are involved.
• Comparative performance evaluation: DoGSite3 enables large-scale comparative performance evaluation on published data sets.

Methodology:

DoGSite3 employs a grid-based method using a Difference-of-Gaussian filter to detect cavities on the protein surface, with a normalization procedure applied before binding site prediction to achieve rotation and translation invariance.