'1-4 pairs were missing' warning when parsing the converted topology with ParmED

[meneshail]

Hi,

I am using acpype to convert my amber topologies to GROMACS format with the command acpype -x <.rst7> -p <.prmtop> -d -a amber. The conversion runs successfully.

But when I tried to parse the generated GROMACS topology file using ParmED with commands like parmed.load_file(top_file, xyz = gro_file), the parser warns about 'GromacsWarning: 114447 1-4 pairs were missing from the [ pairs ] section and were set to zero; make sure you know what you're doing!'.

After a manual check of the <.top> file and the source code of the parser, I found that there are exactly 114447 rows of pair interactions under the section [ pairs_nb ], but there is no [ pairs ] section in my converted <.top> file. Therefore, the ParmED parser neglects the [ pairs_nb ] section and finds no [ pairs ] section, which leads to such a warning. After renaming the [ pairs_nb ] to [ pairs ] and rerunning the parser, the warning is no longer prompted.

So my questions are,

1. Is it an intentional behavior to name the 1-4 pair interactions under section [ pairs_nb ], but not [ pairs ]? Will this section of topology still be used in the GROMACS MD engine?
2. If I want ParmED to read my <.top> file correctly, is it necessary to rename the section [ pairs_nb ] to [ pairs ]?

Thanks for providing this convenient and powerful tool! Yuyang

[alanwilter]

I need to check but it could be just GMX renaming things and ACPYPE needs to catch up or ParmEd issue. You can ask about ParmED at AmberMD mailing list.

[meneshail]

Thanks for the speedy reply! After checking your code now I could see what's happening. There should have been a normal [ pairs ] section, but since I have nonuniform 1-4 scale factors from the GLYCAM06 force field, an additional patch is implemented as below: https://github.com/alanwilter/acpype/blob/f0bdfe831ac82fee4569bb9834a0d26b84293560/acpype/topol.py#L205-L222

The patch basically does these things:

1. Write a new pair interaction section with self.print_gmx_pairs() (this will be the [ pairs_nb ] section)
2. Find where the original [ pairs ] section is in the buffer.
3. Modify the first row of <.top>, setting gen-pairs from 'yes' to 'no'
4. Skip the original [ pairs ] section, and insert the new [ pairs_nb ] at the same place

So if I don't get it wrong, the replacement of the [ pairs ] section with the [ pairs_nb ] section is pretty much an intentional operation? The use of [ pairs_nb ] is because we need to pass different scale factors to different interactions, and the removal of [ pairs ] section is because we don't want the interactions to be computed twice?

[alanwilter]

Indeed, it was intentional.

The details are here:

BERNARDI, A., FALLER, R., REITH, D., and KIRSCHNER, K. N. ACPYPE update for nonuniform 1–4 scale factors: Conversion of the GLYCAM06 force field from AMBER to GROMACS. SoftwareX 10 (2019), 100241. Doi: 10.1016/j.softx.2019.100241

[meneshail]

I see. Thanks again!