Re: [gmx-users] Virtual site and constraints

[Нилов Дмитрий]

I want to present my problem once more.
I'm going to simulate formate with 4fs step size. 
This is formate molecule:
          H

          C

      O1     O2
So I try to remove bond-angle vibration defining three constraints
(H-O1, H-O2, O1-O2). And also I try to keep my molecule planar
defining carbon as massless virtual site type3.
My problem is how to keep the right moment of inertia 
for my molecule when I define carbon like massless.
In this case David van der Spoel advises me to define additional
virtual sites, but virual site could not have mass. 
So I suppose that I should use additional "dummy mass" or 
additional usual atom with zero nonbonded parameters,
but I don`t know how to define its position in molecule
(perhaps I should use some constraints for it?).

Could you help me, please?
----------------
Nilov Dmitri

-----Original Message-----
> Virutal particles should not have any mass.
> Its, e.g. for placing a delocalized charge (as in TIP4P) somewhere.
> How about carefully reading the warnings and try to do, what they suggest?
> How about this?:
>  >      5   opls_966      1    FMT      D      1        12.011    0.0
> 
> Now you have a charge there and no mass. Thats, how virtual sites work.
> If you want to use a "free energy dummy", you may have to use a 
> different approach.
> 
> Regards
> 
> Maik Goette, Dipl. Biol.
> 
> Нилов Дмитрий wrote:
> > I`ve tried to name it like atom (instead of "dummy mass")
> > in this way:
> > 
> > 1)enter new atom type in "ffoplsaa.atp":
> >  opls_966   12.01100  ; Virtual site in formate
> > 
> > 2)enter description of new atom type in "ffoplsaanb.itp":
> > ; name      bond_type  mass    charge  ptype  sigma     epsilon
> >   opls_966  DF      0  12.011  0.0000  D      0.00e+00  0.00e+00
> > (also I`ve tried to use "A" instead of "D" in ptype column)
> > 
> > 3)correct my topology:
> > [ atoms ]
> > ;   nr       type  resnr residue  atom   cgnr     charge       mass
> >      1   opls_279      1    FMT      H      1       0.22      1.008
> >      2   opls_271      1    FMT      C      1       0.58      0.000
> >      3   opls_272      1    FMT     O1      1       -0.9    15.9994
> >      4   opls_272      1    FMT     O2      1       -0.9    15.9994
> >      5   opls_966      1    FMT      D      1        0.0     12.011
> > 
> > [ constraints ]
> > ;  ai    aj funct    c0
> >     1     3     2    0.19850
> >     1     4     2    0.19850
> >     3     4     2    0.22400
> > 
> > [ virtual_sites3 ]
> > ; Dummy from funct a b
> > 2 1 3 4 1 0.345 0.345
> > 5       1       3       4       1       0.345   0.345
> > 
> > But it does not work because grompp runs with error
> > "virtual site D (Res FMT-1) has non-zero mass 12.011".
> > I suppose, that only "dummy masses"(particles named like "M*") could
> > have mass.
> > Could you help me, please?
> > --------------------
> > Nilov Dmitri  
> > 
> > -----Original Message-----
> >> Нилов Дмитрий wrote:
> >>> OK, could I add "dummy mass" that has coordinates and 
> >>> mass such as carbon to keep right moment of inertia?
> >>> I mean something like this:
> >>> [ atoms ]
> >>> ;   nr       type  resnr residue  atom   cgnr     charge       mass  
> >>>      1   opls_279      1    FMT      H      1       0.22      1.008   
> >>>      2   opls_271      1    FMT      C      1       0.58      0.000   
> >>>      3   opls_272      1    FMT     O1      1       -0.9    15.9994   
> >>>      4   opls_272      1    FMT     O2      1       -0.9    15.9994   
> >>>      5         MW      1    FMT      D      1        0.0     12.011
> >>>
> >>> Should I use some constraints for "dummy mass" to define its position
> >>> in simulation? And can I get "dummy mass" type from ffoplaa.atp or
> >>> its better to add the new one?
> >> Have you tried naming it an atom instead of a virtual site?
> >>> I`m sorry, I`ve read manual and corresponding article but I have not
> >>> find irrefragable answer for my questions.
> >>>
> >>> Thanks
> >>> ------------
> >>> Nilov Dmitri
> >>>
> >>>  
> >>>> Нилов Дмитрий wrote:
> >>>>> Thank you very much, but in what way can I define the carbon
> >>>>> like massless when it treated as a virtual site(type 3)
> >>>>> in formate molecule?
> >>>> You can define it as zero, but take care to get the right moment of 
> >>>> inertia (by defining further virtual sites).
> >>>>
> >>>>> Grompp runs with error when carbon has mass:
> >>>>>
> >>>>> converting bonded parameters...
> >>>>> #  CONNBONDS:   3
> >>>>> #     CONSTR:   3
> >>>>> #     VSITE3:   1
> >>>>> Setting particle type to V for virtual sites
> >>>>> ERROR 2 [file "fmt.top", line 82]:
> >>>>>   virtual site C (Res FMT-1) has non-zero mass 12.011
> >>>>> -----------------------------------------------------       
> >>>>>
> >>>>> -----Original Message-----
> >>>>>> Нилов Дмитрий wrote:
> >>>>>>> Hello!
> >>>>>>> I would like to involve formate molecule (HCOO-)in my dynamics, but I have the problem when I use virtual site (type 3fd)for it.
> >>>>>>>
> >>>>>>> When I run grompp for single formate molecule, it`s ok.
> >>>>>>> But after solvating in bath of solvent (tip3p) grompp runs with error:
> >>>>>> In this case you would be best off defining three constraints from
> >>>>>> H O1
> >>>>>> H O2
> >>>>>> O1 O2
> >>>>>> and define the carbon as a virtual site defined by the plane (check 
> >>>>>> TIP4P topology).
> >>>> -- 
> >>>> David van der Spoel, Ph.D.