[gmx-users] Porting CHARMM topologies and parameters

[Mark Abraham]

David van der Spoel wrote:
> Jerome Henin wrote:
> 
>> On Wednesday 16 August 2006 23:43, Mark Abraham wrote:
>>
>>> Jerome Henin wrote:
>>>
>>>> Hi all,
>>>>
>>>> As quite a few people have done before, I am about to try and convert a
>>>> CHARMM topology into a GROMACS one. To that effect, I will probably 
>>>> need
>>>> one or several programs that will do part of the job. When these are
>>>> done, I suppose they might be of interest to others on this list.
>>>>
>>>> Since I am not keen on the idea of reinventing the wheel, I would 
>>>> like to
>>>> know if there is an existing base I can build on. Have you guys been
>>>> doing something of the sort? Using what tools?
>>>
>>> I have previously made available under GPL two perl scripts; one that
>>> convert CHARMM parameter files into GROMACS parameter files (.itp), and
>>> another that corrects the output of grompp to achieve correct treatment
>>> of U-B terms, 1,4 LJ terms and multiple periodic dihedrals. The former
>>> now works on a range of CHARMM parameter files. I attempted to update
>>> the link on the GROMACS web page today, and it has disappeared,
>>> presumably awaiting moderation. Please mail me off-list for a updated
>>> copy if they don't appear in the next few days. There's no change to
>>> their function, but the former script now accept a wider range of 
>>> inputs.
>>>
>>> For residue topology (.rtf) files, you need to consult those of Yuguang
>>> Mu, available here
>>> http://www.gromacs.org/old/topologies/uploaded_force_fields/charmm_gromacs. 
>>>
>>> tar.gz
>>
>>
>> Thanks a lot for the info! I will try that. From what I see, there 
>> might be some work left on the residue topology conversion.
 >>
>>>> A side question: does is sound reasonable to emulate Urey-Bradley angle
>>>> potentials by adding an extra "bond" term?
>>>
>>> U-B angle potentials are implemented natively in GROMACS - see chapters
>>> four and five in the manual. There are of course a few ways you could
>>> hack an emulation of them.
>>
>>
>> I have a problem with that, though. I did look in the manual, and 
>> Urey-Bradley potentials are mentioned in chapter 4, but I cannot see 
>> any reference to them in chapter 5. For instance, table 5.4 lists all 
>> bonded interactions, and none of the 3 angle potentials there is 
>> Urey-Bradley. But it seems that combining a harmonic angle with a 
>> harmonic potential (bonds, type 6) is a clean way of obtaining that 
>> result.
>> Have I missed something?
>>
>> Thanks,
>> Jerome
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please don't post (un)subscribe requests to the list. Use the www 
>> interface or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> 
> it's there. angle type 4 or 5, try it and check your tpr file. sorry 
> that the documentation is a bit outdated.

It's 5, actually, there are cross-bond-bond and cross-bond-angle 
functions at 3 and 4, neither of which make it into Table 5.4. You can 
see this working in the source code of src/gmxlib/ifunc.c.

The advantage of actually using this function type is that you get a 
separate energy term for U-B, rather than having it split over a 
harmonic angle and a harmonic distance potential, (which will have other 
contributions in the former case) which is much better for testing the 
implementation against another CHARMM implementation. It'd be moot for 
production calculation, however.

Mark