[gmx-users] Charge Group Number in itp File (cgnr)

[Mark Abraham]

On 11/09/2012 7:55 AM, Andrew DeYoung wrote:
> Hi,
>
> In the [ atoms ] section of an .itp file, there is a column (cgnr) where a
> charge group number is given (an integer).  This cgnr column is shown in,
> for example, the manual version 4.5.4 in Section 5.7.
>
> My question is, do the charge group numbers have to be _consecutive_
> integers?

No, where relevant they need only differ from the previous group. A 
casual look at the code suggests that consecutive atoms listed with 
charge groups 1 2 1 2 1 2 respectively will end up 6 different charge 
groups.

> For some reason, I just noticed after running many simulations that the .itp
> file for one of my molecules has a missing charge group number.  For
> example, the cgnr column includes the series 425, 426, 428, ... .  That is,
> for some very odd reason (I must have made a typo), there is no charge group
> with number 427.  Will this cause inaccuracies in the simulation?  I do not
> want to rerun all the simulations unless it seems like the missing integer
> will cause problems.

I think it is highly unlikely. You can inspect the .tpr with gmxdump, 
search for "cg" and have a look at the charge group distribution. You 
can also construct the .itp "correctly" and re-process with grompp and 
demonstrate with gmxcheck that both .tpr files are identical.

Mark