fww4f-LigandScoringProtocol.txt

Name

FBRHGRM Method 2

Software

Open Babel 2.3.2/Surflex-Sim 2.745/AutoDock Vina 1.1.2/Clusterizer 1.0/DockAccessor 1.0

Parameters

Assumed pH 7.4

Method

3D PDB structures of the not experimentally D3R compounds have been obtained using the Open Babel suite,
and pre-aligned to the experimentally resolved FXR agonists (available from the Protein Data Bank website and released from the D3R after the end of Stage 1) using Surflex-Sim, to generate
an improved starting point for the refined ensemble docking (since the D3R's 36 FXR protein structure have been included in the original ensemble of protein conformations). MGLTools was used to convert PDBs into PDBQTs.
The grid box was determined according to the crystallized keys and D3R's FXR ligands. AutoDock Vina have been used to run the docking simulations.
The Clusterizer & DockAccessor programs have been used to extrapolate the lowest energy (D3R-not experimentally resolved)ligand-protein conformations.

For each D3R (not experimentally resolved) FXR ligands, the predicted lowest energy estimated by AutoDock Vina is submitted with this protocol.
For the 36 released D3R's FXR agonists (experimentally resolved), the scores have been derived using AutoDock Vina while considering the relative energy minimized co-crystal structures.
D3R crystallographic ligands with double occupancies: the final predicted binding energy of each ligand with multiple occupancies has been derived by weighting (according with the occupancy ratio) the relative binding energies predicted from each experimental ligand pose.

IMPORTANT: for each of the D3R crystallographic ligand-protein complex with double occupancies, all the multiple complexes selected to derive the final binding energy are included:
the multiple poses are placed in the SuppInfo folder. In the SuppInfo folder is also included a csv file (All_affinities_including_considered_multiple_occupancies.csv) reporting all the predicted energies (also for the multiple considered poses) and the calculation applied to derive the corrected energy from the considered multiple structures.

fww4f-PosePredictionProtocol.txt

Name

FBRHGRM Method 2

Software

Open Babel 2.3.2/UCSF Chimera, alpha version 1.11 (build 40585)/GROMACS 5.0.2/MODELLER 9.14/Surflex-Sim 2.745/AutoDock Vina 1.1.2/Clusterizer 1.0/DockAccessor 1.0

System Preparation Parameters

Crystallization pH #used to derive each published and provided D3R's co-crystal structure (for the minimization process)
AMBER ff99SB-ILDN charges #for the minimization process of published and provided D3R's co-crystal structure
Crystallization pH #used to derive each D3R co-crystal structure (for the docking simulations)
Gasteiger charges #for the docking simulations of not experimentally resolved D3R ligands

System Preparation Method

For preparing the locks, all the human FXR co-crystal structures available from the Protein Data Bank
and released by the D3R after Stage 1 have been cleaned (removal of water molecules, etc..).
The non-terminal missing residues were rebuilt using MODELLER 9.14 while maintaining the experimentally-determined portion fixed;
for each complex, 50 models were built and the structure with the lowest DOPE score was selected.
The obtained complexes have been protonated according to each relative crystallization pH (different co-crystals have been crystallized
at different pH) and energetically minimized in a water box using the GROMACS 5.0.2 program with the AMBER99SB-ILDN force field.
All the water molecules were removed after the minimization.
The minimized proteins have been isolated from the relative co-crystallized ligands and used as locks to perform a subsequent ensemble docking.
3D PDB structures of the not experimentally resolved D3R compounds have been obtained using the Open Babel suite
and pre-aligned to the experimentally resolved FXR agonists to generate an improved starting point for the refined ensemble docking.
MGLTools was used to convert PDBs into PDBQTs. The grid box was determined according to the crystallized keys and D3R's FXR ligands.

Pose Prediction Parameters

A.starting_model = 1 #index of the first model for MODELLER
A.ending_model = 50 #index of the last model for MODELLER
Integrator = steep #algorithm (steep = steepest descent minimization) for GROMACS
Emtol = 1000.0 #stop minimization when the maximum force < 10.0 kJ/mol for GROMACS
Nsteps = 5000 #maximum number of (minimization) steps to perform for GROMACS
Exhaustiveness =100 #exhaustiveness of global search for AutoDock Vina
Vina scoring function
Num_modes = 20 #max number of poses to generate for AutoDock Vina
Energy_range = 10 #energy difference (kcal/mol) between the best and worst binding mode for AutoDock Vina

Pose Prediction Method

Docking runs were executed with the above specified parameters.
The Clusterizer & DockAccessor programs have been used to extrapolate the lowest energy (D3R)ligand-protein conformations.
The top (lowest energy) poses from the Vina docking is submitted with this protocol.

IMPORTANT: for each of the D3R crystallographic ligand-protein complex with double occupancies all the multiple complexes selected to derive the final binding energy are included:
the multiple poses are placed in the SuppInfo folder.