Ligand_shape_similarity_docking
AutoDock Vina 1.1.2/RDKit 2018.03.1/Openbabel/Pymol/fkcombu/ACPYPE/Gromacs 5.1.5
Default parameters used for most programs.
3D structure of target compounds was generated using RDKit from smiles.
Hydrogen was added and conformation was generated with ETKDG method in RDKit.
Openbabel was used to assign charges and atomtypes to ligand as PDBQT files.
MGLTools was used to generate PDBQT files for receptor protein.
Vina docking generated 10 poses, with flexible residues within 5 angstrom of the ligand.
The ligand force field parameters for gromacs was generated by ACPYPE with bcc charge method.
Vina scoring funtion (empirical + knowledge-based function)
Publically available structures of Beta-secretase 1 from pdb database was obtained.
The ligands were extracted and compared with target compound using Maximum Common Substructure method in RKDit.
The most similar ligand and its corresponding protein structure A was used as template.
The 20 target compounds were flexibly aligned and superimposed by fkcombu software with the most similar ligand as template.
The target compound was docked into protein A using AutoDock Vina with flexible residues within 5 angstrom of the ligand.
At most 10 poses were generated. The best pose was manually picked.
The protein-ligand complex was energy-minimized using Gromacs.
Yes
Vina_dock_ngf3d_score
AutoDock Vina 1.1.2/RDKit 2018.03.1/Openbabel/Pymol/fkcombu/ACPYPE/Gromacs 5.1.5/ngf3d score function
default
The target compounds for affinity ranking were docked into protein which has the most similar ligand, as described in pose prediction protocol.
After energy minimization, Ngf3d score function was used to predict the binding affinity.
Ngf3d was a home-made score function combining molecular graph representation and deep neural network.
Yes