[gmx-users] How to increase the number of parameters in dihedral angle potentials?

[Mark Abraham]

bhong at Princeton.EDU wrote:
> Dear all,
> 
> This is the problem I meet with. GROMACS provides a Ryckaert-Bellemans function with 6 parameters (C0, C1, C2, ..., C5) , while what I need is a function of summation over 8 terms, that is, 8 parameters from C0 to C7. I tried to add the values of  the two additional pamameters at the end of each lines in the section of dihedral angles in the 'molecule.itp' file. The 'grompp' command then gave an error information, saying there could only be either 6 or 12 parameters for Ryckaert-Bellemans potentials. 

If you read the description of R-B in the manual, it only allows for 
6-parameter R-B. It's not obvious (see footnote to table 5.4, at least), 
but the other are for free energy calculations that might be varying 
these parameters.

>So I added 4 zeros for the rest 4 parameters in 'molecule.itp'. Now, the 'grompp' command did not gave error information any more and I run the simulation with a 12-parameter Ryckaert-Bellemans potential. The problem is that the trajectory obtained from 12-parameter RB potential is exactly the same as the trajectory having generated by 6-parameter RB potential. That means the newly added 6 parameter values can not be recognized by GROMACS. Has anyone met with such kind of problems? 

Hence, your observations. Since you're not running FE, the 7th and 8th 
are never being used for anything.

>What did you do to make GROMACS recognize 6 more parameters in RB potentials? 

You can't. You would have to either implement your own bonded function, 
or use a table lookup for this bonded interaction. See manual section 4.2.13

Mark