[gmx-users] Use different fudgeLJ and fudgeQQ values in simulations with the AMBER force field

[chris.neale at utoronto.ca]

Stephane:

The problem is the coulombic 1-4 interactions. There is not (and never  
has been as far as I can tell) a way to get different 1-4 coulombic  
scaling for different pairs. There is a trick that works when  
combining one ff that uses fudgeQQ 1.0 and another that uses fudgeQQ  
0.5:  
http://oldwww.gromacs.org/pipermail/gmx-users/2010-March/049428.html .

This trick can be applied to obtain micelle fudgeLJ 1.0 and fudgeQQ  
1.0 and, simultaneously, peptide fudgeLJ 0.5 and fudgeQQ = 5/6 (close  
but not exactly 0.8333 due to rounding) as follows:

1. set the fudgeLJ to 1.0 and fudgeQQ to 0.16666666666666

2. calculate the micelle [ pairtypes ] interactions according to the  
combination rules of the ff and then divide the values by 6. You can  
include these "values/6" in the pairtypes section or directly in the  
pairs section.

3. calculate the protein [ pairtypes ] interactions according to the  
combination rules of the ff and then divide the values by 10. You can  
include these "values/12" in the pairtypes section or directly in the  
pairs section.

4. Include all of the micelle [ pairs ] 6 times and all of the peptide  
[ pairs ] 5 times in their respective topologies

The energy should be nearly correct. The only problems will be (a)  
different internal rounding due to different order of operations (this  
will be very small and will get even smaller in double precision) and  
(b) based on the fact that 0.8333 is not exactly 5/6 in the first  
place (you're stuck with this one).

You will need to test that I got the values correctly above. To do  
that, you can compute the pair energy of micelle and protein systems  
individually using the standard ff files and then again using the  
modified version that is detailed above and that should work for both.

If you use this, an appropriate reference would be as an extension of  
the half-epsilon double-pairlist method:
Biophys J. 2010 Mar 3;98(5):784-792. "An Iris-Like
Mechanism of Pore Dilation in the CorA Magnesium Transport System."
Chakrabarti N, Neale C, Payandeh J, Pai EF, Pomès R.

Chris.

-- original message --

Dear gmx users,

Recently I have parametrized a new force field for glycolipid molecules
based on the GLYCAM parameters. I would like to use it with the
AMBER99SB-ILDN force field in gromacs for simulations of peptides and
glycolipds. As you know GLYCAM uses two different values for fudgeLJ and
fudgeQQ (e.g. set to 1.0) instead of 0.5 and 0.8333 values. So it is
possible to define two different fudgeLJ and fudgeQQ values in GROMACS
(as in AMBER11, for example) for the micelle and the peptides.

I have found an old response (2007) in the list archive for a old
version of gromacs

http://lists.gromacs.org/pipermail/gmx-users/2007-February/025907.html

Since this response have been posted 4 years ago, i am not sure that
this approach can be also use with the current version of gromacs ?

Thank you in advance for your response


Stephane