CANDOCK

CANDOCK performs hierarchical atomic network-based flexible small-molecule docking to reconstruct ligands from an atomic grid and sample biologically relevant ligand conformations while accounting for protein flexibility, solvent effects, metal ions, and cofactor interactions for binding-mode prediction and scoring.

Key Features:

• Hierarchical reconstruction: Reconstructs ligands from an atomic grid using a hierarchical approach to generate candidate poses.
• Graph-theory reconstruction: Utilizes graph theory to guide ligand reconstruction from atomic fragments.
• Generalized statistical potential functions: Applies generalized statistical potential functions for conformational sampling and scoring.
• Protein flexibility modeling: Accounts for protein flexibility during docking to capture conformational variability of the binding pocket.
• Explicit chemical environment: Incorporates solvent effects, metal ions, and cofactor interactions within the binding pocket.
• Biologically relevant sampling: Samples conformational spaces aimed at reproducing biologically relevant ligand conformations and binding modes.
• Selector and ranker potentials: Identifies optimal selector and ranker potential functions to correlate statistical scores with experimental binding affinities.
• Benchmark evaluation: Validated on benchmark datasets including PDBbind, Astex, and PINC proteins to assess binding-mode reproduction and scoring performance.

Scientific Applications:

• Structure-based drug design: Predicts ligand binding modes and informs structure-based lead optimization.
• Binding-mode prediction: Reproduces ligand binding poses independent of initial conformations for target characterization.
• Scoring-function assessment: Evaluates and selects statistical potential functions that correlate docked-pose scores with experimental binding affinities.
• Modeling metal/cofactor interactions: Studies ligand interactions in binding pockets that include metal ions and cofactors.

Methodology:

Reconstructs ligands from an atomic grid using graph theory, applies generalized statistical potential functions for sampling and scoring, employs a hierarchical approach to generate conformations independent of initial ligand conformations, identifies selector and ranker potential functions to correlate statistical scores with experimental binding affinities, and evaluates performance on PDBbind, Astex, and PINC proteins.