Too far hydrogen generated from "charmmgui corrected" pdb?

[jdandrade-gmx]

Dear developers,

I came across this situation, and I'm not absolutely certain it is an issue with pdb2pqr, but I thought it would be nice to inform.

It happened when preparing a protein (pdb id: 5v5z) for docking in autodock. With ADT, I can verify that the pqr file created has a hydrogen atom on ILE25 residue that is located too far (so much that AutoDock Tools show it loosely in space), however the "charmmgui corrected" (to include missing residues) pdb file has this hydrogen in a proper distance.

This was reproduced by other users in autodock mail list.

Is this an issue on pdb2pqr, or somewhere else (including with this user)?

[sobolevnrm]

Hello --

I don't observe this problem in VMD:

Can you provide more information about how to recreate using other tools? I don't have access to ADT.

Thank you

[rwxayheee]

Hi @sobolevnrm, Thanks a lot for your kind reply. I talked to @jdandrade-gmx on the AutoDock server. Here's a bit more detail on how I reproduced the unusual bond distance:

I used the PDB2PQR web service. With a provided structure from @jdandrade-gmx, all hydrogen removed, I had a slightly elongated NH bond (1.3 angstrom) in the generated PQR. ADT and other distance-based topology builders might read this as an unbound hydrogen if the distance is out of the permitted bond length.

I feel this might have something to do with the debump or other functions PDB2PQR uses to optimize hydrogens, and it is specific to the input structure. @jdandrade-gmx might be willing to share the input structure and how the model was built. In my generated PQR, there is also a close contact (1.1 angstrom) between two hydrogens, one from the NH group of ILE 25 and the other from the NH group of LEU 26.

The problem could be avoided if this protein was modeled differently. In this PDB2PQR job (https://server.poissonboltzmann.org/jobstatus?jobtype=pdb2pqr&jobid=owrx0e81xv&date=2023-08-01) the AlphaFold model (https://alphafold.ebi.ac.uk/entry/P10613) was used as the input structure.

Hope the information is mildly helpful. Thank you for your time and kind support as always.

Bests,

[jdandrade-gmx]

Hi @sobolevnrm

As @rwxayheee mentioned, the problem arises from the following procedures:

5V5Z structure is pipelined throgh charmm-gui and its lacking residues are added (this creates the "5v5z_corrected.pdb" file in the zip archive, which does not have this issue).

This corrected file is then provided to PDB2PQR service to properly add hydrogens at the pH 7 (this creates the "5v5z_corrected.pqr" file in the zip archive, which does have this issue).

5v5z_corrected.zip