About parameters of pseudoatoms CGn and Gn for further Autogrid4

[songyingit]

Hello,

Recently I read the paper on chemrxiv (https://doi.org/10.26434/chemrxiv-2022-txdt0). It's an amazing work so I also want to try Autodock-GPU for my peptide systems. But, after the .gpf file was generated, when I use autogrid4 to generate the map.fld file, I met this error:

ERROR:  unknown ligand atom type G0
add parameters for it to the parameter library first!

I think this happened because no parameters of pseudoatoms CGn and Gn in parameter library of Autogrid. Therefore, may I ask about where could I get the parameters for these pseudoatoms CGn and Gn?

Thanks, Song

[diogomart]

Hello, I suspect we might have a problem with our documentation regarding autogrid and the .gpf file. How did you produce the .gpf file?

[diogomart]

AD-GPU and Vina do not require maps for those types, so there's no need to compute them with autogrid. This script may help.

[songyingit]

Hello, thanks for the reply!

I used the prepare_receptor4.py in AutodockTools to prepare the .pdbqt file for receptor And then used mk_prepare_ligand.py in Meeko to prepare the .pdbqt file for ligand (also use this flag: "-r C=C-C=A -b 2 3" to disable the rotation of conjugated bonds based on the paper) Then I use prepare_gpf4.py in AutodockTools to produce .gpf file. The command is prepare_gpf4.py -l ligand.pdbqt -r protein.pdbqt -y It works until I try to use Autogrid4 to produce the map.fld file. The command I used here is autogrid4 -p protein.gpf. It showed the error:

ERROR:  unknown ligand atom type G0
add parameters for it to the parameter library first!

I think it should be the same case for CG0, which is another pseudoatom.

BTW, previously, when I used to prepare_ligand4.py in AutodockTools to prepare the .pdbqt file for ligand, it doesn't show any error when I try to get the map.fld file. But when I use this to run Autodock-GPU, it couldn't predict any samples. I think it may be because the peptide I am docking is large (~19 residues), and there are more than 20 rotatable bonds. That's why I want to follow this paper's process to use Meeko to take care of preparing ligand file.

I appreciate if more details could be discussed.

Thanks, Song

[diogomart]

Thank you for explaining your steps thoroughly. prepare_gpf4.py considers only the atom types in the ligand, but usually it's more efficient to run autogrid once for all atom types that may occur in ligands, enabling docking of any ligand without further autogrid calls. This is what the other script does. It will help in your case specifically because the CGn/Gn atom types are not added to the GPF. Eventually, we'll probably stop recommending prepare_gpf4.py and use the other script instead.

[songyingit]

Thank you so much for the script. I have tried this script to prepare .gpf file for my systems, and it could work now.

Best, Song