[gmx-users] 1,4 interactions

[Justin A. Lemkul]

Gavin Melaugh wrote:
> Right, I think the penny has finally dropped. On the back of what you said.
> 
> 1)In my topologies I have always listed say a [bond] directive, with
> bond indices,function, k, and r0. I could however just list the indices
> and the function, and then have the corresponding parameters in the
> [bond types] directive, which save me associating the parameters to each
> individual bond ?
> 

Yes.

> 2) In reponse to what yo said finally "If you list specific pairs of
> atoms in a [pairs] directive, then grompp will search for the
> appropriate parameters listed in a [pairstypes] directive.  If none are
> found, and "gen-pairs = yes," grompp generates the appropriate
> parameters for you." Are the parameters generated from the [atomtypes]
> directive?
> 

Yes, in conjunction with FudgeLJ and FudgeQQ.

-Justin

> Gavin
> 
> Justin A. Lemkul wrote:
>>
>> Gavin Melaugh wrote:
>>> A few points then,
>>>
>>> 1) So a [pairs] directive would include all the atom indices of the 1,4
>>> interactions and the coressponding values of sigma and epsilon?, which I
>>> would do manually.
>>>
>> No.  A [pairs] directive simply lists the pairs of atoms for which 1,4
>> interactions should be calculated.  No parameters are listed here. 
>> Format:
>>
>> atom_i atom_j func_type
>>
>>> 2) In the manual it says ''gen-pairs = yes : generate 1,4 interactions
>>> which are not present in the pair list from normal LJ parameters using
>>> Fudge LJ". Sorry if I am coming across stupid, but that seems different
>>> to what you're saying?
>>>
>> Not at all.  The parameters for a specific interaction are given in a
>> [pairtypes] directive.  If these parameters are not present and
>> "gen-pairs = yes" then they are calculated.  If parameters for a
>> specific pair are not present, and "gen-pairs = no" then grompp
>> generates a fatal error.
>>
>>> 3) Also am Icorrect about the normal nb-LJ coming from the [atom types
>>> directive]
>>>
>> Yes.
>>
>>> 4) I suppose a very obvious question would be; What is the difference
>>> between a [pairtypes] directive and [pairs] directive (bonds,
>>> bondtypes etc)
>>>
>> A [*type] directive lists the parameters that should be a applied to
>> whatever interaction type might be found.  For instance, a [bondtypes]
>> directive contains default parameters for all the bonds you might
>> find, like C-C.  If you then have a [bonds] directive that lists:
>>
>> ; ai  aj  func
>>    1   2    1
>>
>> ...and atoms 1 and 2 are both C, then grompp finds the corresponding
>> parameters in the [bondtypes] directive and applies them to that bond,
>> a process which it carries out for all the rest of the interaction
>> types, including pairs.  If you list specific pairs of atoms in a
>> [pairs] directive, then grompp will search for the appropriate
>> parameters listed in a [pairstypes] directive.  If none are found, and
>> "gen-pairs = yes," grompp generates the appropriate parameters for you.
>>
>> -Justin
>>
>>> Gavin
>>>
>>> P.S I appreciate you using your time to explain this
>>> Justin A. Lemkul wrote:
>>>> Gavin Melaugh wrote:
>>>>> Hi Justin
>>>>>
>>>>> Thanks for replying. I have checked the energy file and there are
>>>>> no 1,4
>>>>> terms, which is what I expected. I should clarify that my
>>>>> parameters are
>>>>> taken from the OPLS ua forcefield but I have relabelled them for my
>>>>> own
>>>>> model, which is not a common molecule. Therefore all the info on the
>>>>> molecule is in .top and obviously the .gro file. i.e. there are no
>>>>> #include statements. Now I take it that the non-bonded LJ are taken
>>>>> from
>>>>> the [atom types] directive? The question is then how and where do I
>>>>> state the 1,4 terms. I always thought that when gen-pairs = yes, then
>>>>> gromacs would generate a list of all the 1,4 interaction types?
>>>>>
>>>> You don't explicitly state 1-4 terms with OPLS-AA.  While you don't
>>>> need a [pairtypes] directive, you do need a [pairs] directive to
>>>> dictate the 1-4 interactions.  Setting "gen-pairs = yes" generates the
>>>> parameters, such that a [pairtypes] directive is unnecessary.
>>>>
>>>> -Justin
>>>>
>>>>> Gavin
>>>>>
>>>>> Justin A. Lemkul wrote:
>>>>>> Gavin Melaugh wrote:
>>>>>>> Hi all
>>>>>>>
>>>>>>> Can someone please get back to me on this. I have generated my own
>>>>>>> topology file from OPLS parameters by hand with the following
>>>>>>> format. I
>>>>>>> am now concerned that I haven't been calculating 1,4 interactions,
>>>>>>> which
>>>>>>> I thought were generated by setting gen-pairs = yes. I have read
>>>>>>> section
>>>>>>> 5 of the manual several times but am still unclear about this.
>>>>>>>
>>>>>> You can answer this question by analyzing your .edr file: are there
>>>>>> LJ-14 and/or Coul-14 terms?  If so, then you're calculating 1-4
>>>>>> interactions.  Seems to me that, for OPLS, your approach is fine.  It
>>>>>> wasn't clear from your earlier posts that you were using OPLS, hence
>>>>>> the inability to know if what you were doing was right or not.
>>>>>> Different force fields have different underlying mechanics, so the
>>>>>> use
>>>>>> of [nonbond_params] and/or [pairs]/[pairtypes] is somewhat dependent
>>>>>> upon your model.
>>>>>>
>>>>>> -Justin
>>>>>>
>>>>>>> Cheers
>>>>>>>
>>>>>>> Gavin
>>>>>>>
>>>>>>> parameter level
>>>>>>>> [defaults]
>>>>>>>> ; nbfunc    comb-rule     gen-pairs        fudgeLJ     fudgeQQ
>>>>>>>>      1         3              yes            0.5         0.5
>>>>>>>>
>>>>>>>> [atomtypes]
>>>>>>>> ;type     mass           charge      ptype     sigma(nm)  
>>>>>>>> epsilon(kjmol-1)
>>>>>>>>
>>>>>>>>
>>>>>>>> ;Molecular level
>>>>>>>> [moleculetype]
>>>>>>>> ;       name         nrexcl
>>>>>>>>         tertpentane      3
>>>>>>>>
>>>>>>>>
>>>>>>>> [atoms]
>>>>>>>> ; atomnr  type   resnr  residue   name    cgnr     charge     mass
>>>>>>>>
>>>>>>>>
>>>>>>>> [bonds]
>>>>>>>> ;  ai    aj   func   b0(nm)       kb(kjmol-1 nm-2)
>>>>>>>>
>>>>>>>>
>>>>>>>> [angles]
>>>>>>>> ;  ai    aj    ak   func  theta(deg)    ktheta(kjmol-1 rad-2)
>>>>>>>>
>>>>>>>>
>>>>>>>> [dihedrals]
>>>>>>>>
>>>>>>>>
>>>>>>>> ;System level
>>>>>>>> [system]
>>>>>>>>
>>>>>>>>
>>>>>>>> [molecules]
>>>>>>>> ; moleculename     number
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>   
>>>
> 
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================