QM/MMGBSA charge error

[Alexandre Pinto]

Hi,
I've been trying to run QM/MMGBSA with the following command:
gmx_MMPBSA -O -i qmmmgbsa.in -cs em3_1_7.pdb -ci index.ndx -cg 0 1 -ct em3_1_7.xtc -cp PROCESSED.top  -o FINAL_RESULTS_MMPBSA.dat -eo FINAL_RESULTS_MMPBSA.csv

However, I can't define the qm charges correctly... I have this in the input:
 qmcharge_rec=0,
 qmcharge_lig=-1,
 qmcharge_com=-1,

But I get this in the ouput:
R:A:GLY:88  R:A:SER:89  R:A:GLY:90  R:A:THR:91  R:A:GLY:181 R:A:CYS:182 R:A:SER:183 R:A:GLU:184 R:A:ARG:187 R:A:ALA:207
R:A:LEU:212 R:A:TRP:355 R:A:ARG:369 R:A:PHE:373 R:A:SER:374 R:A:SER:377 R:A:HIE:446 R:A:SER:449 R:A:ASP:450 R:A:PRO:451
R:A:VAL:452 R:A:PHE:453 R:A:ASN:485 R:A:HID:486 R:A:CYS:487 L:B:MOL:561

[WARNING] Setting qmcharge_com = -1.0
[WARNING] Setting qmcharge_rec = -1.0
[WARNING] Setting qmcharge_lig = 0.0

[Mario Sergio Valdes]

Thanks for reporting it.

Previously, it was common for users to forget to assign the charge to the selected residue groups, thus ending in an error. It was also frequent that they arbitrarily assigned charges, which is not entirely correct. We decided to compute the charge during the selection process from the topology, eliminating the two most common sources of error. However, reviewing the code from this notification, I noticed that the user-defined value is not applied, but instead is forced to use the computed value. I have already fixed it. Please check if it works for you.

python -m pip install git+https://github.com/Valdes-Tresanco-MS/gmx_MMPBSA.git -U

[Alexandre Pinto]

Dear Mario,

I confirm that the user-defined charges for the ligand and complex are now being read instead, but the receptor charge is still being automatically defined. 

R:A:GLY:88  R:A:SER:89  R:A:GLY:90  R:A:THR:91  R:A:GLY:181 R:A:CYS:182 R:A:SER:183 R:A:GLU:184 R:A:ARG:187 R:A:ALA:207
R:A:LEU:212 R:A:TRP:355 R:A:ARG:369 R:A:PHE:373 R:A:SER:374 R:A:SER:377 R:A:HIE:446 R:A:SER:449 R:A:ASP:450 R:A:PRO:451
R:A:VAL:452 R:A:PHE:453 R:A:ASN:485 R:A:HID:486 R:A:CYS:487 L:B:MOL:561

[WARNING] Using user-defined qmcharge_com. Are you sure of this value?

[WARNING] Setting qmcharge_rec = -1.0

[WARNING] Using user-defined qmcharge_lig. Are you sure of this value?
[WARNING] Defined qmcharge_lig (-1) is different from the computed value (0.0). Are you really sure of this value?

I'm wondering if my topology in its current form is not compatible with gmx_MMPBSA, since the charge of the receptor should be 0 (according to the number of positive and negative residues) and the charge of the ligand -1, and they are being automatically defined as -1 and 0, respectively.

My best regards,

Alexandre Pinto

[Mario Sergio Valdes]

I'm wondering if my topology in its current form is not compatible with gmx_MMPBSA, since the charge of the receptor should be 0 (according to the number of positive and negative residues) and the charge of the ligand -1, and they are being automatically defined as -1 and 0, respectively.

This charge is defined for the selected residues only not for the receptor. We calculate it automatically because the charges are already defined in the topology. In this sense, I don't think it's a good idea to force the program with new charges because you could get artificial results.

In any case, I already solved (I think) the problem, but I have not been able to test it. Please test it and tell me if it works for you.

Mario S.

[Alexandre Pinto]

Dear Mario,

Now I can run the QM/MMPBSA calculation with all the charges being user-defined, but I'm not sure if I can trust the results, since charges not consistent with the system will lead to artificial results like you said. 

With the automatic charges gmx_MMPBSA aborts when the calculation gets to the receptor. I'm not sure I understand why the automatic charges are being defined like this... Shouldn't the receptor charge only include the receptor QM residues, as the ligand is absent in the amber files related to this part of the calculation?

My best regards,

Alexandre Pinto

[marioe911116]

Hi Alexandre! Could you please attach the files you are using to take a closer look? we don't have that many system to test QM/MMGBSA and this way will be good to improve the processing for these kind of calculations

thanks in advance

[Alexandre Pinto]

Dear Mario,

I'm sending the files attached bellow.

Best regards,

Alexandre Pinto

[Mario Sergio Valdes]

Hi Alexandre.

Sorry for the delay. I was traveling and doing immigration procedures.
I already identified the error. You were correct. I had not identified the problem because I did not have the structure. The error occurs when you select the residues manually instead of by distance (as in the example). The sum of charges was inverted. I fixed it. Please check if it works for you and let me know.

Best

Mario S.