Failure at Fragments #84
Description
Hi Poltype Team,
I am attempting to use poltype2 to create parameters for a small ligand molecule. In my initial attempt, I left most options related to fragmenting as default, i.e. dontfrag=False, fragbigmultipole=True, etc. During the run, 5 fragments were created, with only 2 of which had Poltype Job Finished shown in the log.
One of the fragments had a convergence failure in the first optimization run. Checking the structure shown in Gaussian input .com file I found the fragment to be chemically unreasonable, where a carbon originally in a carboxylate group was included but without the two oxygens, as well as having no extra hydrogen capped. Essentially it is single bonded to another carbon, while having no other bonds. I would think this could be resolved if I include maxgrowthcycles to be 6 or 8, so the two oxygens originally there will be included to the structure too, but I'm not sure if missing capped hydrogens there is a bug or not.
2 other fragments ended with "Value Error: Energy is greater than 55 kcal/mol." Both structure looks chemically reasonable, thus I wasn't sure where to proceed from there. Do you have any suggestions on what I can do here?
In a separate attempt, I set dontfrag=True, and the poltype2 job finished without failing, creating a final.key and final.xyz. However, I noticed in the README output that the torsion fitting was a bit poor:
Fri Sep 12 22:39:52 2025 Absolute or Relative RMSPD of QM and MM torsion profiles is high, RMSPD = 1.3928509429988873 Tolerance is 1 kcal/mol RMSPDRel =0.27324605407464636 tolerance is 0.1
Fri Sep 12 22:39:52 2025 Absolute or Relative RMSPD of QM and MM torsion profiles is high, RMSPD = 1.9070179049289078 Tolerance is 1 kcal/mol RMSPDRel =0.6641889633455356 tolerance is 0.1
Fri Sep 12 22:46:35 2025 Warning: RMSD of QM and MM optimized structures is high, RMSD = 1.348179 Tolerance is 1
Fri Sep 12 22:46:35 2025 Relative error of 0.05150427138789159 for QMDipole 8.077 and 7.661 for MMDipole tolerance = 0.5 /
In some of the Torsion Plots, the QM and MM2 post-fist torsion curves also looked not quite fitted in some parts of the curve. Is there anything I can do or check to improve this?
Thank you very much for your time.