ICM Dock
Molsoft ICM 3.8-7b
standard
Standard ICM conversion procedure was used to add hydrogens, assign atom types and charges.
All compounds were docked to an ensemble of 2 PDBs (4DPF) and 3 structures released in Stage 1b (BACE_2, BACE_7, BACE_14)
For each protein structure, two alternative protonation states were generated in which either Asp 80 or Asp 276 was protonated.
14 molecules from PDBs (2F3E, 2F3F, 3DUY, 3DV1, 3DV5, 3K5C, 3K5D, 3K5G, 4DPF, 4DPI, 4GMI, 4K8S, 4KE0, 4KE1),
and the 20 co-crystallized ligands released in Stage 1b were used as Atomic Property Field bias.
For each protein structure, a docking energy offset was generated by redocking 34 co-crystallized molecules, optimizing the ligand and receptor Calpha RMSD
Monte Carlo search 'effort' parameter was set to 10.
Docking runs were executed with the above specified parameters while default values
were applied for the rest of the variables. Three independent docking runs were performed and resulting poses combined.
The top pose according to the sum of ICM VLS score and APF similarity to the closest template was selected
No
Yes
ICM Dock - APF 3D QSAR
Molsoft ICM 3.8-7b
For pose prediction Monte Carlo search 'effort' parameter was set to 10.
Standard ICM conversion procedure was used to add hydrogens, assign atom types and charges.
All compounds were docked to an ensemble of 7 receptor conformations from PDB (2F3E, 2F3F, 3K5C, 3K5D, 4DPI, 4GMI, 4PDF)
For each PDB, two alternative protonation states were generated in which either Asp 80 or Asp 276 was protonated.
14 molecules from PDBs (2F3E, 2F3F, 3DUY, 3DV1, 3DV5, 3K5C, 3K5D, 3K5G, 4DPF, 4DPI, 4GMI, 4K8S, 4KE0, 4KE1) were used as Atomic Property Field bias.
For each PDB, a docking energy offset was generated by redocking 14 co-crystallized molecules, optimizing the ligand and receptor Calpha RMSD
Docking runs were executed with the above specified parameters while default values
were applied for the rest of the variables. Three independent docking runs were performed and resulting poses combined.
The top pose according to the sum of ICM VLS score and APF similarity to the closest template was selected
1397 Compounds from ChEMBL_24 were selected, they were docked using the same procedure.
Their 3D poses were used to train a 3D-QSAR model to predict their pKd value.
The 3D-QSAR model was then applied to the challenge compounds, the predicted pKd value is converted into Binding Energy.
No
Yes