[gmx-users] Re: Creating an atomtype where all nonbonded interactions are enumerated

[Andrew DeYoung]

Justin,

Thanks.  So I imagine that the order that the atoms are "stepped through" is
dictated by the order of atoms in the configuration file (.gro or .pdb); do
you think that this is correct?  I am not simulating a protein,  so other
than the order of atoms in the configuration file, for my system there is
not any obvious order to use (such as N terminus, H attached to N terminus,
C-alpha, etc).

Thanks!

Andrew DeYoung
Carnegie Mellon University

> I haven't dissected this information for a while, but here's my
understanding. 
> Someone please correct me if I'm wrong.  The list contains all the
function 
> types that are used in the simulation, with each assigned an index.  What
you're 
> interested in are the LJ_SR parameters, which are constructed (again, this
is 
> just my interpretation from systems that I have used) by stepping through
all of 
> the atoms in each molecule and adding a new function when it is needed.
That 
> is, for a simple protein using Gromos96 53A6, the first few lines are (my 
> comments added after ';'):

> functype[0]=LJ_SR, c6= 2.43640970e-03, c12= 2.31952890e-06  ; NL-NL
> functype[1]=LJ_SR, c6= 0.00000000e+00, c12= 0.00000000e+00  ; H-NL
> functype[2]=LJ_SR, c6= 3.84514406e-03, c12= 1.50015503e-05  ; CH1-NL
> functype[3]=LJ_SR, c6= 4.26569115e-03, c12= 8.87604438e-06  ; CH2-NL

> Thus it is stepping through each new atomtype (mapped to N, H, C-alpha,
C-beta, 
> etc) in the amino acid and generating parameters for interactions between
itself 
> and all prior atomtypes that have been encountered.  You can check in the 
> [nonbond_params] directive to see where some special values are used.  For

> Gromos96, NL-NL interactions have no special values to them, the normal
C6/C12 
> parameters are used.  This is not, however, the case for other atomtypes,
where special combinations are used.