3DLigandSite

3DLigandSite predicts ligand binding sites in proteins by mapping ligands from structurally similar proteins onto query structural models to infer probable binding locations.

Key Features:

• Prediction methodology: Employs a methodology rooted in manual approaches used during CASP8 and uses structural models generated from protein-structure predictions to identify potential ligand binding sites.
• Comparative modeling and ligand superposition: Identifies structurally similar proteins and superimposes their bound ligands onto the query model to transfer ligand positions.
• Input types: Accepts query sequence or query structure as input for generating structural models and predictions.
• Use of experimentally determined ligands: Compares query models with known structures that have experimentally determined ligands to infer binding-site locations.
• Benchmarking performance: Achieved a Matthew's correlation coefficient (MCC) of 0.64 on CASP8 targets with coverage of 71% and accuracy of 60%, and an MCC of 0.68 on a larger dataset of protein structures.

Scientific Applications:

• Drug Discovery: Predicts potential ligand binding sites to help identify novel targets and guide small-molecule design.
• Protein Engineering: Informs design and modification of proteins by revealing likely ligand interaction sites.
• Structural Biology: Assists in interpreting protein-ligand interactions by transferring experimentally observed ligand positions from homologous structures.

Methodology:

Generate structural models for target proteins using protein-structure prediction techniques; identify structurally similar proteins with experimentally determined ligands; align and superimpose those ligands onto the query model to predict ligand binding sites, following approaches used in CASP8.