CHARMM Development Project
I want to derive parameters for Ser-PLP (see attachment). For this first, I tried CGenFF web site and got a really high penalty. Also, it failed to identify the Imine bond (C=N). And gave N atom type of NG311. I have used PDB as well as mol2 with connectivity information.

With one of the advices I received from an earlier post ( I tried using parameters from analog compounds as much as I can.

For pyridinium, I was able to find parameters from two places. (1) 3-hydroxypyridine and ethylpyridine from toppar_all36_prot_pyridines.str, and also from all36_cgenff.rtf for protonated pyridine (CG2R62). I am planning to derive charges and other parameters for these compounds. However, what atom type would be most reasonable to take out of these two options. In 1st option, I dont have the protonated version. But in the second option I dont have other substituents. Any assistant is highly appreciated.

Attached File
ser_plp.png  (235 downloads)
Note that the .mol2 file must have the correct bond order for the C=N bond, or the CGenFF server may be confused. The H atom placement in the molecule, esp. protonation states, must also be correct.

A high penalty doesn't necessarily mean the estimated parameters are wrong, just that the analogy to existing validated parameters is not very good. It means at least testing the estimated parameters, which should be done for whatever approach is used to get a first test set, CGenFF or chemical intuition. Multiple parameter sets are often tested in simulations, so it is not unreasonable to test both choices and the CGenFF estimates.

It should be noted that QM calculations are encouraged, if possible, when the penalties are very large.

The problem for testing is often the difficulty in finding good target data from experiments that help refine the conformational preferences in the appropriate environment, e.g. water and/or a protein binding site. Aggressive and persistent searching of the literature and other sources for the whole molecule or relevant fragments is often needed to find some useful data.

If the molecule will form an important part of the research project, the decisions made on these parameters are critical. Some additional, unplanned effort may be required in order to ensure producing a respectable result for the parameters.
Thank you very much sir!! I did discover why CGenFF server got confused based on your feedback. I manually adjusted the bond order earlier,and hence added the bond order of 2 for C=N and others accordingly. But I realized I should add "ar" in the bond order column since the pyridine system is a conjugated one. Updated mol2 file gave me CG2R62 for the C and others as I expected.

> Multiple parameter sets are often tested in simulations, so it is not unreasonable to test both choices and the CGenFF estimates.

Okay sir. I will test both types of parameters to evaluate them.

PLS-Ser does play an important role. Because of this, I did design a positive control and a negative control with our experimental collaborators to validate our simulation models.

Hopefully, through that I might be able to ensure the accuracy of the simulation setup (fingers crossed).

Thank you very much for your kind feedback (as always) on this.
© CHARMM forums