Surflex/GRIM Interaction Pattern Similarity/ HYDE (Protocol3)
PROTOSS v2.0/Corina v3.40/Filter v.184.108.40.206/ Sybyl v2.1.1/Surflex-v3065/IChem/HYDE 3.2.5; v.1.3.0.
protonation state of protein assigned according to pH 7.4; the defined ligands` stereocenters in the input file were preserved.
I.Protein preparation: 25 available FXR structures deposited in the PDB were used in the docking protocol (PDB id: 1OSH, 1OSV, 1OT7, 3BEJ, 3DCT, 3DCU, 3FLI, 3FXV, 3GD2, 3HC5, 3HC6, 3L1B, 3OKH, 3OKI, 3OLF, 3OMK, 3OMM, 3OOF, 3OOK, 3P88, 3P89, 3RUT, 3RUU, 3RVF, 4QE6). The co-crystallized inhibitors were removed and all the structures were aligned with the coordinates of the APO structure provided by the Challenge. Hydrogen atoms were added using PROTOSS and protonation of the active site residues was verified manually. A mol2 file was generated for each protein using Sybyl. For one of the PDB templates (1OT7), two chains were considered, since chain B was co-crystallized with two coactivator peptides being referred here as 1OT7_A and 1OT7_B.The water molecules located in a sphere of 6.5 Ã around the co-crystallized ligands were maintained. II.Ligand preparation: The provided SDF files were converted into MOL2 format using CORINA preserving the defined stereo centers in the input file. The correct protonation was assigned with Filter and verified manually. Few corrections in the atom-types and bonds were made when necessary. III.Docking setup: Surflex uses a pseudo-molecule, called protomol, as the target to which align putative ligands of a protein as binding site. The protomol was generated for each one of the 26 PDB structures, based on a list of residues located in a sphere of 6.5 Ã around their corresponding co-crystallized ligand. The options -proto_thresh 0.3 and -proto_bloat 3 were used in order to expand the extent of the protomol. The protein residues are kept rigid. Other parameters were assigned as default.
Surflex uses Morphological similarity function and fast pose generation techniques as search method; Surflex uses the Hammerhead scoring function; 20 conformations generated for each ligand; the -pgeom option ensured search coverage and the returned poses are different from one another by at least 0.5 Angstrom rmsd.
Docking runs were performed, according to the above parameters, while default values were assigned to rest. The poses with Surflex docking score inferior to 2.0 were disconsidered. We docked the 102 FXR ligands, into each one of the selected 25 PDB structures. To re-score the docking poses, we used the GRIM Interaction pattern similarity method, developed in our group and implemented on the software IChem. It allows an interaction-based alignment of different protein-ligand complexes that can be quantified by an empirical scoring function (GRIM score) and used to post-process docking poses by similarity to known protein-ligand interaction patterns, in our case, the selected 25 FXR protein structures used in the docking runs complexed with their co-crystallized ligands. To calculate the GRIM score, IChem requires a .mol2 file containing the active site residues of the protein; a .mol2 file containing a reference ligand (in this case the co-crystallized inhibitors complexed with the PDB structures used in the docking) and a single or multi .mol2 file containing the ligands to be re-scored. Using IChem, we then compared each docking pose to the selected PDB complexes. For each FXR ligand, we selected 5 poses with the highest values of GRIM scores, considering all the PDB complexes used in the comparison. Since we previously discarded the poses with Surflex docking score inferior to 2.0, some ligands have less than 5 predicted poses. For the selected poses, we calculated their binding affinity in kcal/mol using the HYDE scorer. In this protocol, the five poses for each FXR ligand (when available) are ranked by their GRIM score, where 1 indicates the pose with the higher GRIM score value. Nevertheless, we provide in the .mol files, the associated HYDE energy with the GRIM-ranked poses