Research Object Crate for Consensus Virtual Drug Screening Workflow

This workflows contains a pipeline in Scipion that performs the following steps: 1.1) Import small molecules: introduces a set of small molecular structures in the pipeline as prospective ligands 1.2) Import atomic structure: introduces a protein atomic structure in the pipeline as receptor. 2.1) Ligand preparation: uses RDKit to prepare the small molecules optimizing their 3D structure. 2.2) Receptor preparation: uses bioPython to prepare the receptor structure, removing waters, adding hydrogens and removing unnecessary chains if asked. Also, uses PDBFixer to optimize the structure if selected. 3.1) Ligand filters: uses RDKit to perform ADME and PAINS filters on the prepared ligands to remove undesired molecules 3.2) Protein pocket search: uses 3 different software (P2Rank, AutoSite and FPocket) for predicting the receptor pockets. 4.2) Consensus pockets: common pockets are computed by clustering their contact residues in order to obtain the most promising pocket predicted by all 3 programs. 5) Receptor-ligands docking: uses 3 different software (AutoDock-GPU, AutoDock-Vina and LeDock) to dock the prepared ligands onto the receptor pockets. 6) Docked poses rescoring: uses ODDT Vina scoring to rescore the poses coming from all 3 different software in order to have a comparable score of the poses. 7.1) Consensus docking: common ligand poses are computed clustering by RMSD the different molecules in order to obtain the most promising predicted poses. 7.2) Ranx scoring: the scores of the different programs are combined using Ranx in order to obtain a final score for each of the molecules.

Author

License

Apache-2.0

Contents