I compare the GROMACS carbohydrate topologies for the AMBER-compatible GLYCAM-06j force field generated by the two programs:
doGlycans
(Reinis Danne, Chetan Poojari, Hector Martinez-Seara, Sami Rissanen, Fabio Lolicato, Tomasz Róg, and Ilpo Vattulainen. doGlycans–Tools for Preparing Carbohydrate Structures for Atomistic Simulations of Glycoproteins, Glycolipids, and Carbohydrate Polymers for GROMACS
Journal of Chemical Information and Modeling 2017 57 (10), 2401-2406 DOI: 10.1021/acs.jcim.7b00237)
and ACPYPE- processed of topology generated in the Amber tool tLeap
(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).
Both topologies take into account the features of 1-4 interactions of the force field of the GLYCAM-06j, but this is done in different ways.
While doGlycans uses the [pairs] section and the function 2 to explicitly indicate the Lennard-Jones interaction scale factor = 1,
acpipe uses section [pairs_nb] function 1, while disabling generation of gen-pairs = no in the topology in section [defaults].
Can anyone explain to me the similarities and differences of such approaches, their influence on MD simulation and which one is more correct?
Generate pair parameters allows to avoid explicitly typing any that follow combination rules. This reduces the number of written parameters if many follow combination rules and avoid typing errors. But this gives the risk of forgetting to add pair parameters that do not follow combination rules. This means one could easily do simulations with incorrect parameters.
Not using parameter generation simply requires all 1-4 parameters to be supplied. No risk of incorrectly using combined parameters, but there is, of course, the risk of typos.
When used correctly they are identical and correct.
Berk, thank you very much, I thought about that, but I did not understand enough the difference between the sections [pairs] and [pairs_nb]. It seems to me that the ACPYPE method is more risky and will require explicit indication of all interaction parameters 1-4 in a mixed topology of carbohydrates, for example, with proteins, which will greatly increase its size. How ACPYPE will do this, I will check, maybe GLYCAM forcefield is sufficient for protein parameters as well…
Hi, I have essentially done the same thing: generating a topology (or itp) using doglycans and acpype.
Personally, I see the use of ACPYPE over doglycans because I like to parameterize my systems in amber with tleap before converting to gromacs. However, doglycans seems more user friendly and straight forward when building carbohydrates and if you have the forcefield libraries you use downloaded in gromacs then that’s great.
I get the same results where ACPYPE gives [pairs_nb] and function 1 and doglycans gives [pairs] and function 2 with a fudgeLJ of 1.
In this case as the scale factor is 1 and my V and W values that doglycans and ACPYPE give are equal.
However, what I do not fully understand is if in the [defaults] section I had a fudgeLJ defined as 0.5 for example: would this affect the V and W values in the [pairtypes] section, or because they are explicitly given the fudgeLJ scaling factor is not applied?
This is important to know for me because I want to be sure that my parameters for my carbohydrates are not being re-scaled when they are in a topology file where I have [default] settings with fudge values given for my protein.