100x10M_noxraywat_adadelta_drylig_yesfilter_yesbias
OpenCLAutoDock/AutoDock4.2/OpenBabel/Python/Reduce
Assumed pH 7.4, single protonation state
Gasteiger charges for both ligand and protein
Macrocycles broken
Ligands protonated at standard pH using OpenBabel based scripts.
Only one protonation state is considered. The large macrocyclic substructures of ligands
are broken to allow conformational sampling during the docking.
Proteins protonated with reduce, single protonation state.
X-ray waters are removed from proteins.
100 Genetic Algorithm (GA) instances
10 million energy evals per GA
Local search frequency: 100 percent
Local search max evals: 500
Local search method: Adadelta
Ligands were docked onto 11 protein structures chosen to
cover multiple conformations of the binding pocket
(2b8v, 2f3f, 2p4j, 2wf3, 3l59, 3msj, 3msk, 4weo, 4fs4, 4rcf, 5hu0).
Macrocyclic ligands were docked in an open conformation and
a linear potential was used to bring the broken bond back together (50 kcal/mol/angstrom).
The scoring function was biased to penalize docking poses that differ from the x-ray
coordinates of ligand 4dpf, with respect to atoms O15, N16 and C18.
The penalty was applied to atoms further than 1.2 angstroms from the corresponding
reference atom (either O15, N16 or C18 in 4dpf).
Every pose was subjected to post-processing filters to evaluate similarity with known
binding modes - we refer to these filters as substructure filters.
The following smarts strings and associated reference ligands were considered:
4dpf, [CX4][CX4]([OH1])[CX4][NX3][CX3]=O
2f3f, [CX4][NX3][CX3](=[O])[CX4]([CX4])[CX4][CX4]([OH1])[CX4][NX3][CX3]=O
4dpf, [CX4][N][CX4][CX4]([OH1])[CX4][NX3][CX3]=O
4dpf, [CX4][c][c][c][CX4][N][CX4][CX4]([OH1])[CX4][NX3][CX3]=O
4dpf, [CX4]([OH1])[CX4][NX3][CX3](=O)[a]1[a][a][a][a][a]1
2f3f, [CX4]([OH1])[CX4][NX3][CX3](=O)[CX4][NX3][CX3](=[O])[CX4]
4k8s, [CX4]([OH1])[CX4]([CX4][a]1[a][a][a][a][a]1)[NX3][CX3](=O)
Among the poses passing the substructure filters, which may have been docked
in any of the 11 receptors, the pose with the
lowest AutoDock4.2 energy is considered as the correct pose.
No
Yes
MM-GBSA
amber18,MMPBSA.py(Amber16)
Assumed pH 4.5
AM1-BCC charges
0.1 M NaCl solution for MD simulations
Amberff99sb, TIP3P and GAFF2 for MD simulations
Amber parameter, dt=0.002,ntc=2,ntf=2,cut=8.0, ntb=2, ntp=1, taup=2.0, ntt=3, gamma_ln=2.0, temp0=300.0
For mmgbsa calculations used MMPBSA.py, igb=8, saltcon=0.100
Ligands were docked according to pose prediction protocol named 100x10M_noxraywat_adadelta_drylig_yesfilter_yesbias.
Structures for the receptors were obtained from Protein Data Bank.
Pdbfixer was used to remove the ligand and the water molecules (if any) and to add the missing heavy atoms.
Then, PDB2PQR server (http://nbcr-222.ucsd.edu/pdb2pqr_2.0.0/) was used to correct the protonation states at pH 4.5
and to fix the residue/atom names following AMBER naming scheme.
Parmed was used to convert the resulting pqr file to a pdb file.
OpenEye toolkits were used to check bond-order and connectivities.
For MM-GBSA calculations, the protein-ligand complex was solvated in TIP3P water in a cubic box with 10 Angstrom padding.
The protein-ligand complex were minimized and simulated for 14 ns using explicit solvent MD simulations in NPT ensemble and the MM-GBSA calculations were performed on the last 10 ns to estimate the binding free energy.
No
No