[gmx-users] Re: prodrg on small molecule

[Pim Schravendijk]

gb_26 is defined in gmx/share/top/ffG43a1bon.itp:

#define gb_26       0.1530  7.1500e+06
; C, CHn  -   C, CHn    800     

as you can see the bond-distance is the same for both, but PRODRG
generated an harmonic bond ("fu" is 1), while in GROMOS 43a1 there is a
GROMOS-96 bond ("fu" is 2). In chapter 4.2 of the manual "bonded
interactions" you see the corresponding equations. You can see there 
that the bond constants of both functions are related as:

k_harm = 2*k_g96*b^2 

if you fill in 7.1500e+06 for k_g96 you get 334749 for k_harm so in effect
both your potentials here have the same force constant, you could choose
to change all your bond/angle/dihedral/nonbonded potentials to the
gromos96 43a1 type without much problems. 

Be aware that partial charges from PRODRG can be not completely correct,
check this with the charges of similar molecules that are listed in
gmx/share/top/ffG43a1.rtp

hope this answers your question!

greetings, Pim

> 
> Maybe this is a really useless quesiontion, but I really need your help.
> Thank you in advance!
> 
> I use PRODRG Beta2 to generate GORMOS96 force field on my small molecule,
> but finally I found the format is still like Gromacs force field.
> 
> PRODRG Beta2 generated the results as following:
>  [ bonds ]
>  ;ai  aj  fu    c0          c1
>    1   2   1 0.153    334720.0 0.153    334720.0 ;   C18  C13
>    2   3   1 0.153    334720.0 0.153    334720.0 ;   C13  C12
> .............
> 
> 
> But in GROMOS96(43a1), should be like this:
> [ bonds ]
>  ;ai  aj  fu    c0          c1
>      1     2     2 gb_26
>      2     3     2 gb_26
> ............
> 
> So my question is should I change all the format as those in
> GROMOS96(43a1) style? If so, some of the bonds are not defined in
> GROMOS96(43a1), so I should add these missing bonds into it manually,
> right?
> 
> Thank you for any reply.
> 
> Jinzhi
> 2005-10-17
--
Pim Schravendijk - PhD Student
Max Planck Institute for Polymer Research
http://www.mpip-mainz.mpg.de/~schraven/