Computational docking of small-molecule ligands into proteins is an important tool for modern drug discovery. Despite that some conformational differences are observed between free and ligand-bound states, protein conformational flexibility is often ignored in protein-small molecule docking programs. Our integrative, information-driven docking approach HADDOCK [1,2], which does account for flexibility in both receptor and ligand, has been mainly applied to protein-protein, protein-DNA/RNA, and protein-peptide docking problems. Here we present the results of its application to a benchmark of 76 protein-ligand complexes [3], concentrating on unbound cases for 56 of those native complexes, using an optimized ligand-specific protocol.<br/>Overall HADDOCK performance (top-scoring cluster with ligand-RMSD &lt; 2Å) on the benchmark set of Native cases (redocking) was over 79% using ligand-specific protocols, and generated a top5-scoring solution in over 92% of all 76 Native cases.<br/>For the Unbound cases (each Native case represented by an apo-structure and/or a structure crystallized with a different ligand), the overall success of HADDOCK was up to 69%, and a top5-scoring solution in up to 83% of the Unbound cases. The HADDOCK Unbound docking performance is relatively insensitive to conformational differences in the input protein, and was successful for interface-RMSDs up to 3.0Å compared to the Native binding site.

Information-driven fully flexible HADDOCKing: Performance on a benchmark of protein-ligand complexes

253rd National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.