[gmx-developers] additional rtp entries

[Mark Abraham]

On 29/09/2011 1:06 AM, Alexey Shvetsov wrote:
> On Wed, 28 Sep 2011 23:03:34 +1000, Mark Abraham wrote:
>> On 28/09/2011 8:29 PM, Alexey Shvetsov wrote:
>>> Hi all!
>>>
>>> Also why not make some common directory under $GMXDATA/top for such 
>>> things as cofactors? for example amrbe99* series forcefields has 
>>> compatible atom definitions for cofactors same for amber94 based 
>>> ones. May be this will work for contributed cofactor topology?
>>
>> How would you make it easy for users to know that they exist and with
>> which force field they should be used?
> They can be in directory with name like ligands.amber99 so they can be 
> included in rtp entry search while running pdb2gmx. I can made patches 
> to realize so.

If these .rtp entries are only valid for one force field, then they 
should go in its .rtp or nowhere. If they're valid for more than one 
force field, some mapping must exist between an .rtp entry and its legal 
force fields. Why is writing machinery in pdb2gmx to manage this mapping 
preferable to requiring a user to obtain a custom amber99XX.ff/ tarball 
from a contributions section?

>>
>> The real problem is that anything in the repository has to have
>> someone prepared to take responsibility for checking, documenting and
>> maintaining it, and developers' time is finite. I think a
>> contributions section is a good place for such content. Now users can
>> clearly understand that the level of support implied is distinct, and
>> a web search will find it.
> Yes contribution page is good thing (and it contain some variants of 
> forcefields and itps for limited set of ligands) but there should be 
> easy way to include new rtp entryes in ff. 

There is a reasonably straightforward way to include new entries in a 
force field - see 
http://www.gromacs.org/Documentation/How-tos/Adding_a_Residue_to_a_Force_Field. 
I do not yet see the benefits of new rtp entries in the main GROMACS 
repository. We don't have the resources to curate a universal library of 
useful stuff.

Mark

> Something like i mentioned above (e.g. directory like 
> ligand.<ffbasename> in $GMXDATA/top or in current dir for pdb2gmx)
>>
>> Mark
>>
>>>
>>> On Mon, 19 Sep 2011 16:20:34 +0200, Berk Hess wrote:
>>>> Hi all,
>>>>
>>>> We don't have a strict official policy for inclusion of rtp building
>>>> blocks in the standard
>>>> share/top forcefield directories in Gromacs.
>>>>
>>>> One option would be to only allow "official" force field building
>>>> blocks (where the definition
>>>> of official is already difficult) and put everything else on the
>>>> gromacs.org contributions page.
>>>>
>>>> I bring this up because of a contribution of ATP for AMBER on gerrit:
>>>> https://gerrit.gromacs.org/#change,98
>>>> In this contribution certainly references should be added so it's
>>>> clear where the parameters
>>>> come from. But such contributions should be carefully checked on 
>>>> correctness,
>>>> but equally important, on compatibility with the force field and
>>>> something like "acceptance"
>>>> in the community.
>>>>
>>>> ATP would indeed by useful, but if it's easy to find on the
>>>> contributions page that would
>>>> also work.
>>>>
>>>> Berk
>>>
>