[Question]: Positive Binding Energy

[laiyii]

My Question is...

How can I avoid calculating ΔEDISPER term when performing a CHARMMFF pbsa calculation? I found a solution: https://groups.google.com/g/gmx_mmpbsa/c/pYeE20pbhnc?pli=1 however, I still don't know how to discard this term just by changing 'inp=2' to 'inp=1' in _GMXMMPBSA_info

[Valdes-Tresanco-MS]

Simply change the value of the variable to INPUT['inp'] = 1 in the _GMXMMPBSA_info file and then run

gmx_MMPBSA --rewrite-output

[laiyii]

thanks for your reply!

[laiyii]

I find two ways of discarding ΔEDISPER term (add inp=1 in mmpbsa.in or rewrite output) return different ddG results, up to 60 kcal/mol, it is confusing for me, how is this happen?

[Valdes-Tresanco-MS]

When you add inp=1 to the mmpbsa.in file does the calculation run again? mmpbsa.in is a parameter file for the calculation, so any modification requires rerunning the calculation. On the other hand, the _GMXMMPBSA_info file contains the run configuration, so you can only modify parameters that do not require recalculation, just rewrite the output.

[laiyii]

yes, I did run again.

[laiyii]

cmd: gmx_MMPBSA -O -i mmpbsa.in -cs ../md.tpr -ci ../index.ndx -cg 20 19 -ct ../mono28_%s.xtc -cp ../topol.top -o mono28_%s.dat -eo mono28_%s.csv -nogui mmpbsa.in is bellow: &general sys_name="Prot-Lig-CHARMM", startframe=1, endframe=800, PBRadii=5, / &pb istrng=0.15, fillratio=4.0, radiopt=0,inp=1 /

[Valdes-Tresanco-MS]

Oh, I see what you mean. In itself, both solutions are not equivalent because of the way the non-polar component of the solvation is computed. If the variable inp=1 in the mmpbsa.in file, you must also define use_sav=0 as well. In such a case, the total non-polar solvation free energy will be modeled as a single term linearly proportional to the solvent-accessible surface area. On the other hand, if you modify the INPUT['inp'] variable in the _GMXMMPBSA_info file you will be ignoring the EDISPER term, but the computation of the non-polar component of the solvation, specifically the cavity term will still be maintained having been computed using the molecular volume enclosed by SASA instead of being modeled as a single term linearly proportional to the SASA. Definitely, we must clarify this since it is an approximation in itself so as not to have to recalculate, but it can be confusing and contain errors. We will review this content as soon as possible. Thank you for reporting it.

[laiyii]

thanks for your reply, very glad to help you.

[Valdes-Tresanco-MS]

For the moment, it is best to modify the variables of the input file and rerun the calculation to avoid inaccuracies.

[Valdes-Tresanco-MS]

@marioernestovaldes Please help me with this theory doubt, should we remove the rewrite output option with inp=1? I think the rerun is inaccurate because of what I explained above. Also we must configure the change of use_sav = 0 when inp = 1.

[marioernestovaldes]

clarified inp variable usage in fa0d1e9

[laiyii]

adding inp=1 in mmpbsa.in and both inp=1, use_sav=0 in .in file give same results.

[Valdes-Tresanco-MS]

@marioernestovaldes Reviewing the theory and this section, I think we need to work on improving this. inp = 2 is optimized to use multiple predefined parameters, among them Tan & Luo's radii which are not done. In the Tan & Luo publication, they specify various combinations of optimized parameters depending on the decomposition scheme. These are the things we have to check/do:

• [ ] Tan & Luo radii implementation. Currently, when the radiiopt is set to 0, the radii defined by the user is used. Although it may be common, we should show a warning and clarify it in the documentation.
• [ ] Implement a variable that defines the parameter sets defined by Tan & Luo. That is each value of this variable changes all the associated parameters.
• [ ] Add a warning in the documentation for inp = 1. When defined this way, Sander automatically resets the variables sprob, cavity_surften, and cavity_offset, whose default values are optimized for inp = 2. Should we add more options to this variable?
• [ ] Since the ENPOLAR changes when using inp = 2 or inp = 1, we should remove the option to do the rewrite-output. According to my tests, you have to modify many parameters to get close values, so what we currently do is "inaccurate".

[Clare-xiong]

I have the same issue. When I add inp=1, use_sav=0 in the mmpbsa.in, the results is -6.63, but when I change inp=1 in the _GMXMMPBSA_info and then run gmx_MMPBSA --rewrite-output, the result is -34.99. I want to know which result is more accurate?

[marioernestovaldes]

@Clare-xiong The more accurate value will be the one obtained with the correct parameters in the input file mmpbsa.in as stated here (https://github.com/Valdes-Tresanco-MS/gmx_MMPBSA/issues/273#issuecomment-1207144247). The option of --rewrite-output can be used for relative comparisons though as the "error" is the same for all the systems.

[Clare-xiong]

Thanks for your reply.