D3RGC2-Score-2
CACTVS Chemoinformatics Toolkit v3.433/CORINA v3.60/UCSF Chimera v1.10.2/GOLD v5.2/MODELLER v9.16
Gold software with ASP scoring function
Search efficiency 200% # more adapted for flexible ligands
10 poses generated for each ligand
If necessary, the protein structure was completed using MODELLER and the closest available structure as template. All proteins were superimposed onto the APO structure using UCSF Chimera, so that all docking calculations can be carried out in a coordinates system compatible with the submission. Protein was prepared (hydrogens added) using the Hermes graphical interface of GOLD software. Protein was prepared (hydrogens added) using the Hermes graphical interface of GOLD software. 3D coordinates of the ligands (MOL2 format) were generated from the SMILES format (provided) using CORINA. When applicable, the corresponding enantiomers were also generated. Binding site was defined as a sphere with 20 Å radius around C alpha of Ala288. All FXR crystal structures available were retrieved from the PDB and, together with the APO FXR structure provided by the organizers, were clustered into 6 groups according to their 3D conformational similarity. A representative structure was selected for each of the 6 groups. Re-docking and cross-docking calculations were carried out with all FXR ligands available in the PDB and the 6 representative structures selected previously, using AutoDock, Vina, and Gold with four scoring functions (GoldScore, ChemScore, ChemPLP, ASP). Rigid docking and different combinations of flexible side chains (up to 8) were considered, but for technical reasons only the rigid docking results are reported here. The analysis of these data showed that Gold with the ASP scoring function provided the best results (GoldScore also behaved quite well). AutoDock was not able to position the ligand within the binding site in most of the cases, and therefore it was not further considered. Docking calculations were then carried out with the 102 FXR ligands dataset provided and the 6 representative FXR structures selected previously, using Vina and Gold with four scoring functions (GoldScore, ChemScore, ChemPLP, ASP). Docking runs were executed with the above specified parameters while default values were applied for the rest of the variables. The 102 ligands from the dataset provided were organized into 7 groups: benzimidazoles, isoxazoles, steroids, FXR_5-like, spiro, sulfonamides and others. Crystal structures containing compounds from the first four groups were available, and the biggest common substructure from each group was used as reference for a partial RMSD calculation for each docking pose. This gave us a rough idea about the correctness of docking poses. The RMSD calculation was realized using an in house developed script based on CACTVS Chemoinformatics Toolkit. The poses from the spiro and sulfonamides groups were visually inspected using UCSF Chimera. Only the 3FLI and the APO structures provided docking poses with a carboxylate group (that is present in most spiro structures and in FXR_101 from the sulfonamides group) interacting with Arg335. This was considered as the correct orientation, since most of the crystalized ligands show the same kind of interaction. Overall, poses obtained with the structure 3OLF were selected for benzimidazoles, with 1OSV for steroids, with 3HC5 for isoxazoles and with 3FLI for all others. From these poses, two submissions (each one containing a Pose and Score component) were generated as follows: in Submission 1, the best ranking poses were selected and submitted directly; in Submission 2 (the present file), the poses with the lowest RMSD compared with the reference fragment were selected, and their corresponding scores submitted - for the cases where no reference fragment was available (spiro, sulfonamides, others), the best ranking poses were submitted instead. Protein structures were converted into PDB format for submission using UCSF Chimera, and the docking poses were converted into MOL format using CORINA (the MOL format corresponds to the SDF output format in CORINA).
D3RGC2-Score-2
CACTVS Chemoinformatics Toolkit v3.433/CORINA v3.60/UCSF Chimera v1.10.2/GOLD v5.2/MODELLER v9.16
Protonation state of ligands at pH 7.0 was considered.
If necessary, the protein structure was completed using MODELLER and the closest available structure as template. All proteins were superimposed onto the APO structure using UCSF Chimera, so that all docking calculations can be carried out in a coordinates system compatible with the submission. Protein was prepared (hydrogens added) using the Hermes graphical interface of GOLD software. Protein was prepared (hydrogens added) using the Hermes graphical interface of GOLD software. 3D coordinates of the ligands (MOL2 format) were generated from the SMILES format (provided) using CORINA. When applicable, the corresponding enantiomers were also generated. Binding site was defined as a sphere with 20 Å radius around C alpha of Ala288.
Gold software with ASP scoring function
Search efficiency 200% # more adapted for flexible ligands
10 poses generated for each ligand
All FXR crystal structures available were retrieved from the PDB and, together with the APO FXR structure provided by the organizers, were clustered into 6 groups according to their 3D conformational similarity. A representative structure was selected for each of the 6 groups. Re-docking and cross-docking calculations were carried out with all FXR ligands available in the PDB and the 6 representative structures selected previously, using AutoDock, Vina, and Gold with four scoring functions (GoldScore, ChemScore, ChemPLP, ASP). Rigid docking and different combinations of flexible side chains (up to 8) were considered, but for technical reasons only the rigid docking results are reported here. The analysis of these data showed that Gold with the ASP scoring function provided the best results (GoldScore also behaved quite well). AutoDock was not able to position the ligand within the binding site in most of the cases, and therefore it was not further considered. Docking calculations were then carried out with the 102 FXR ligands dataset provided and the 6 representative FXR structures selected previously, using Vina and Gold with four scoring functions (GoldScore, ChemScore, ChemPLP, ASP). Docking runs were executed with the above specified parameters while default values were applied for the rest of the variables. The 102 ligands from the dataset provided were organized into 7 groups: benzimidazoles, isoxazoles, steroids, FXR_5-like, spiro, sulfonamides and others. Crystal structures containing compounds from the first four groups were available, and the biggest common substructure from each group was used as reference for a partial RMSD calculation for each docking pose. This gave us a rough idea about the correctness of docking poses. The RMSD calculation was realized using an in house developed script based on CACTVS Chemoinformatics Toolkit. The poses from the spiro and sulfonamides groups were visually inspected using UCSF Chimera. Only the 3FLI and the APO structures provided docking poses with a carboxylate group (that is present in most spiro structures and in FXR_101 from the sulfonamides group) interacting with Arg335. This was considered as the correct orientation, since most of the crystalized ligands show the same kind of interaction. Overall, poses obtained with the structure 3OLF were selected for benzimidazoles, with 1OSV for steroids, with 3HC5 for isoxazoles and with 3FLI for all others. From these poses, two submissions (each one containing a Pose and Score component) were generated as follows: in Submission 1, the best ranking poses were selected and submitted directly; in Submission 2 (the present file), the poses with the lowest RMSD compared with the reference fragment were selected, and their corresponding scores submitted - for the cases where no reference fragment was available (spiro, sulfonamides, others), the best ranking poses were submitted instead. Protein structures were converted into PDB format for submission using UCSF Chimera, and the docking poses were converted into MOL format using CORINA (the MOL format corresponds to the SDF output format in CORINA).