CSM-lig

CSM-lig predicts binding affinities of protein-small molecule complexes to quantify and rank molecular interactions relevant to drug development and molecular recognition.

Key Features:

• Graph-Based Structural Signatures: Utilizes graph-based structural signatures to evaluate shape and chemical complementarity between proteins and small molecules.
• Training and Evaluation: Trained and evaluated on various releases of the PDBbind databases, reporting Pearson correlation coefficients up to 0.86 in 10-fold cross-validation and 0.80 in blind tests.

Scientific Applications:

• Docking Pose Assessment: Assesses docking poses to evaluate how well predicted ligand positions match experimental data.
• Mutation Analysis: Assesses the impact of mutations, including insertions, deletions, and alternative splicing, on protein-small molecule affinity.
• Protein-Compound Recognition Studies: Identifies factors that drive recognition between proteins and compounds.

Methodology:

Uses graph-based structural signatures and was trained and evaluated on various releases of the PDBbind databases using 10-fold cross-validation and blind tests.