[gmx-users] How to edit the cgnr and charge fields of a ITP file with the result of Gaussian?

[Mark Abraham]

Liao Chuan wrote:
>  
> 
> Dear Gromacs users,
> 
> I obtained the ITP file of glycerol molecule from the Dundee PRODRG server
> (http://davapc1.bioch.dundee.ac.uk/programs/prodrg/). However, I heard that
> the charges are not accurate enough. So I estimated the charges of glycerol
> using Gaussian. The input file of Gaussian is gol.gjf, including all atoms
> of a glycerol molecule. The output of Gaussian is selected as follows:
> 
> -------------------------- Gaussian Output
> ----------------------------------------------------------------------------
> -----------
> 
> Mulliken atomic charges:
> 
>               1
> 
>      1  O   -0.677102
> 
>      2  H    0.422534
> 
>      3  C   -0.156687
> 
>      4  H    0.224462
> 
>      5  H    0.169912
> 
>      6  C    0.007200
> 
>      7  H    0.182644
> 
>      8  O   -0.586768
> 
>      9  H    0.435753
> 
>     10  C   -0.171961
> 
>     11  H    0.173527
> 
>     12  H    0.190745
> 
>     13  O   -0.630544
> 
>     14  H    0.416285
> 
>  Sum of Mulliken charges=   0.00000
> 
> -------------------------------------------END of Gaussian Output
> ----------------------------------------------------------------
> 
>  
> 
> ------------------------------------------ ITP file
> ----------------------------------------------------------------------------
> ------
> 
> [ atoms ]
> 
> ;   nr      type  resnr resid  atom  cgnr   charge     mass
> 
>      1        OA     1  GOL     OAB     1   -0.095  15.9994   
> 
>      2        HO     1  GOL     HAA     1    0.043   1.0080   
> 
>      3       CH2     1  GOL     CAA     1    0.052  14.0270   
> 
>      4       CH1     1  GOL     CAC     2    0.078  13.0190   
> 
>      5        OA     1  GOL     OAD     2   -0.114  15.9994   
> 
>      6        HO     1  GOL     HAB     2    0.036   1.0080   
> 
>      7       CH2     1  GOL     CAE     3    0.053  14.0270   
> 
>      8        OA     1  GOL     OAF     3   -0.095  15.9994   
> 
>      9        HO     1  GOL     HAC     3    0.042   1.0080   
> 
> -------------------------------------END of ITP file
> ----------------------------------------------------------------------------
> --
> 
>  
> 
> Then it comes to edit the ITP file. But there are some problems puzzling me.
> 
> Firstly, the ITP file just has 9 atoms ( see above) as the Dundee PRODRG
> server uses united-atom force field which neglects 5 unpolar hydrogen atoms
> while the result of Gaussian has 14 atoms. How should I edit the charges of
> ITP file? Or how could I edit the cgnr and charge fields of the ITP file
> with the Gaussian result? Is it reasonable and acceptable that a charge
> group has a decimal number of charges?
> 
> Secondly, the charge group number of ITP file is 3 whereas the result of
> Gaussian has only one charge group(14 atoms as a whole group). Could the
> Gaussian package calculate the charges of a molecules by first dividing the
> atoms of the molecule into different charge groups? 
> 
> Any help/suggestions will be greatly appreciated!

You should read how the united-atom force field for which you are trying 
to get charges worked out the charges used in their parameterization, 
and do something consistent with that. Presupposing the suitability of 
some other approach is likely doomed to failure.

Mark