Re: [AMBER] how to fill in missing parameters in tleap?

From: Astrid Maaß <astrid.maass.scai.fraunhofer.de>
Date: Fri, 09 Sep 2011 13:24:22 +0200 (CEST)

Hi Yun,

you have outlined the procedure the way I would do it.

You already have identified an "offensive" set of atoms with rdparm. In your example these atoms are
3.H2N :3.N2 :3.C2N :3.O2N. The residue number (":3" in this case) is not relevant, but the atom names follow the leading '@'-character are the ones you need. In glycam04.prep I find these names e.g. for the residue 1YB, where the atom names and types (plus their connectivity) are listed. So you need the atom-type combination "H -N -C -O" (or possibly the reversed pattern; mind the blanks). In fact, for this torsion, glycam04.dat shows the following entry (where, ironically, you want to reverse the correction ;-) ).

H -N -C -O 1 2.00 0.0 -1. Parm94 - Corrected to remove phase shift
              1 -2.50 0.0 2.

Modify as suggested earlier and proceed as before.

Good luck!

 Astrid



----- Ursprüngliche Mail -----
Von: "Yun Shi" <yunshi09.gmail.com>
An: "AMBER Mailing List" <amber.ambermd.org>
Gesendet: Freitag, 9. September 2011 08:47:16
Betreff: Re: [AMBER] how to fill in missing parameters in tleap?

Hi Astrid,

I would be very happy to share.

But before I try this, I just want to make sure if I could just find the
pattern of atomtypes concerned (with negative dihedral force constants as
displayed by rdparm) in Glycam_06g.dat file, modify corresponding parameters
in the way you mentioned before, and then use tleap/xleap to generate prmtop
again, which is subsequently converted by amb2gmx.pl?

Thanks,

Yun

On Thu, Sep 8, 2011 at 11:24 PM, Astrid Maaß <
astrid.maass.scai.fraunhofer.de> wrote:

> Hi Yun,
>
> ----- Ursprüngliche Mail -----
> > Von: "Yun Shi" <yunshi09.gmail.com>
> > An: "AMBER Mailing List" <amber.ambermd.org>
> > Gesendet: Donnerstag, 8. September 2011 22:50:55
> > Betreff: Re: [AMBER] how to fill in missing parameters in tleap?
> >
> > Hi Astrid,
> >
> > Thanks for the advice. The problem in amb2gmx.pl has not been fixed,
> > which
> > still do not recognize negative pK. So what you were saying is to
> > change
> > something like:
> >
> > E 53: -2.500 0.00 2.0 :3.H2N :3.N2 :3.C2N :3.O2N
> > (45,44,46,47)
> >
> > to
> >
> > E 53: 2.500 180.00 2.0 :3.H2N :3.N2 :3.C2N :3.O2N
> > (45,44,46,47)
> >
> > ?
> >
> > And then use amb2gmx.pl to convert?
> >
>
> yes, this is exactly, what I meant.
>
>
> > I also wonder if it's OK to change the ">" to "!=" as in
> > ...........................
> >
> > # get all force constants for each line of a dihedral #
> > my $lines = $i -1 +$numijkl;
> > for(my $j=$i;$j<=$lines;$j++){
> > my $period = abs($pn{$j});
> > if($pk{$j}>0) {
> > $V[$period] = 2*$pk{$j}*$cal/$idivf{$j};
> > }
> >
> > ...........................
> >
> > Regards,
> >
> > Yun
> >
>
> Adressing the problem by fixing the code is certainly better, but since I
> am not much of a programming person, I am not sure, whether the fix you
> suggest below remedies the problem. If you decide to try both variants
> (modified input FF + old code; old FF + modified code), I think the
> resulting gromacs topologies should finally display consistent
> RB-coefficients.
>
> I would appreciate if you could share the outcome of this with us.
>
> Regards,
>
> Astrid
>
>
> >
> > On Wed, Sep 7, 2011 at 12:14 AM, Astrid Maaß <
> > astrid.maass.scai.fraunhofer.de> wrote:
> >
> > > Hi Yun,
> > >
> > >
> > > as per amber file format specifications, the minus indicates that
> > > the
> > > dihedral potential of this particular set of atoms is composed of
> > > more than
> > > one cosine term; this appears to be properly recognized, I had not
> > > troubles
> > > with amb2gmx.pl in this context.
> > >
> > > The numbers in columns 6 - 11 of the gromacs-topology are the
> > > Ryckaert-Bellemans-constants you get AFTER converting the
> > > amber-topology to
> > > the gromacs. Do not worry about the signs in the GROMACS-topology.
> > >
> > > What you should check with rdparm, if you create a topology file
> > > for input
> > > in amb2gmx.pl with tleap/xleap/sleap anyways, whether dihedrals
> > > with
> > > negative force constants are present in the AMBER-topology file.
> > > The rdparm
> > > output is easier to read than the topology file itself and tells
> > > you, which
> > > torsion exactly i.e. which atoms are involved and thus which
> > > pattern of
> > > atomtypes is concerned (you can check the link between an atom's
> > > name and
> > > type in prep or mol2 files).
> > >
> > > If you find suspect torsions/force constants in the rdparm-output
> > > (or the
> > > FF-input.dat itself), create a copy of the FF-input.dat and modify
> > > within
> > > this copy the respective force constants & phase shifts for the
> > > identified
> > > patterns. Then create with the modified FF-input.dat a new set of
> > > top- and
> > > crd-files for conversion with amb2gmx.pl. As I said before, I am
> > > not sure
> > > whether the problem might have been fixed since the time when I
> > > encountered
> > > it, so check, whether the whole fuss is necessary and if the old
> > > and the new
> > > gromacs-topologies are indeed different.
> > >
> > > Good luck!
> > >
> > > Astrid
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > ----- Ursprüngliche Mail -----
> > > Von: "Yun Shi" <yunshi09.gmail.com>
> > > An: "AMBER Mailing List" <amber.ambermd.org>
> > > Gesendet: Dienstag, 6. September 2011 20:22:06
> > > Betreff: Re: [AMBER] how to fill in missing parameters in tleap?
> > >
> > > Hi Astrid,
> > >
> > > I saw something like:
> > >
> > > [ dihedrals ]
> > > ;i j k l func C0 ... C5
> > > 19 18 20 21 3 0.75312 2.25936 0.00000
> > > -3.01248 0.00000 0.00000 ;
> > > 41 1 18 19 3 0.20920 0.62760 0.00000
> > > -0.83680 0.00000 0.00000 ;
> > >
> > > after converting glycam ff-parametized saccharide to gromacs
> > > topology
> > > format. I referred to gromacs manual, and knew that '3' in the
> > > fifth column
> > > indicates the Ryckaert-Bellemans-potential you mentioned. But could
> > > you
> > > tell
> > > me what those numbers in columns 6 to 11 mean? Which is force
> > > constant and
> > > which is phase shift?
> > >
> > > Thanks,
> > >
> > > Yun
> > >
> > >
> > > On Tue, Sep 6, 2011 at 8:21 AM, Yun Shi <yunshi09.gmail.com> wrote:
> > >
> > > > Hi Astrid,
> > > >
> > > > Thank you for the info.
> > > >
> > > > But in addition to negative force constant, there are also some
> > > > negative
> > > > multiplicity for GLYCAM dihedral parameters. So would these
> > > > negative
> > > > multiplicities be a problem for amb2gmx as well?
> > > >
> > > > Thanks,
> > > >
> > > > Yun
> > > >
> > > >
> > > >
> > > > On Tue, Sep 6, 2011 at 2:07 AM, Astrid Maaß <
> > > > astrid.maass.scai.fraunhofer.de> wrote:
> > > >
> > > >> Hi Yun Shi,
> > > >>
> > > >>
> > > >> I used amb2gmx.pl some time ago and if I remember correctly, for
> > > >> torsions, amb2gmx.pl converts the Amber-cosine series in a
> > > >> Ryckaert-Bellemans-potential. This was OK in most of the cases,
> > > >> but when
> > > >> amb2gmx.pl encounters a negative force constant it may not
> > > >> translate
> > > the
> > > >> potential correctly. I do not know, whether this behavior has
> > > >> been fixed
> > > in
> > > >> the meantime, but you could bypass it by using a phase shift of
> > > >> 180
> > > instead
> > > >> of a negative sign in the force constant for the respective
> > > >> term. You
> > > might
> > > >> use rdparm to check whether this difficulty would apply to your
> > > >> system
> > > >> anyways.
> > > >>
> > > >> Hope that helps
> > > >> Astrid
> > > >>
> > > >>
> > > >>
> > > >>
> > > >> ----- Ursprüngliche Mail -----
> > > >> Von: "Yun Shi" <yunshi09.gmail.com>
> > > >> An: "AMBER Mailing List" <amber.ambermd.org>
> > > >> Gesendet: Dienstag, 6. September 2011 01:22:19
> > > >> Betreff: Re: [AMBER] how to fill in missing parameters in tleap?
> > > >>
> > > >> Hi all,
> > > >>
> > > >> For this specific molecule, I wonder if it's possible to
> > > >> manually set
> > > >> parameters for the non-standard parts.
> > > >>
> > > >> I was thinking about using GLYCAM and amber99SB, while avoiding
> > > >> the use
> > > of
> > > >> GAFF force field, with the help of amb2gmx.pl and GROMACS
> > > >> software.
> > > >>
> > > >> 1. Set geometrical parameters and atomic charges to all standard
> > > residues
> > > >> using GLYCAM and 99SB.
> > > >>
> > > >> 2. Derive RESP charges for the modified TRP and thio-glycosidic
> > > >> linkage.
> > > >>
> > > >> 3. With GROMACS topology files, set geometrical parameters for
> > > >> the
> > > >> modified
> > > >> TRP and thio-glycosidic linkage according to 99SB. That is,
> > > >> define atom
> > > >> types, bonds (CT-S), angles (CT-S-CT, CT-CT-S), and dihedrals
> > > >> (X-CT-S-X)
> > > >> with corresponding parameters from 99SB. And for N, H, CA, CB,
> > > >> and
> > > indole
> > > >> parts of the modified TRP, just use whatever geometrical
> > > >> parameters from
> > > >> the
> > > >> standard TRP in 99SB.
> > > >>
> > > >> 4. Combine all these parts in GROMACS format.
> > > >>
> > > >>
> > > >> On Sun, Aug 28, 2011 at 6:12 AM, case <case.biomaps.rutgers.edu>
> > > >> wrote:
> > > >>
> > > >> > On Fri, Aug 26, 2011, Jason Swails wrote:
> > > >> > >
> > > >> > > Use parmchk. If you run the program with no arguments, it
> > > >> > > tells you
> > > >> how
> > > >> > to
> > > >> > > use it. By default, it will search gaff.dat and pull out
> > > >> > > reasonable
> > > >> > > parameters, dumping them to an frcmod file. If there's
> > > >> > > anything it
> > > >> > simply
> > > >> > > *can't* find, it will put a placeholder (all parameters are
> > > >> > > 0) and
> > > say
> > > >> > > "ATTN, need revision". You have to fix those parameters.
> > > >> >
> > > >> > Parmchk does a bit more than just look in gaff.dat: it
> > > >> > estimates
> > > missing
> > > >> > parameters by several algorithms. In its default mode,
> > > >> > parameters in
> > > >> > gaff.dat
> > > >> > are not put into the frcmod file; parameters it estimates on
> > > >> > its own
> > > are
> > > >> > output; finally, things it can't figure out get placeholders.
> > > >> > >
> > > >> > > However, I would warn you that gaff dihedrals should not
> > > >> > > necessarily
> > > >> be
> > > >> > > accepted. Run short simulations to make sure they make
> > > >> > > sense, as
> > > you
> > > >> may
> > > >> > > have to modify some of them by hand.
> > > >> >
> > > >> > This is excellent advice.
> > > >> >
> > > >> > ...dac
> > > >> >
> > > >> >
> > > >> > _______________________________________________
> > > >> > AMBER mailing list
> > > >> > AMBER.ambermd.org
> > > >> > http://lists.ambermd.org/mailman/listinfo/amber
> > > >> >
> > > >> _______________________________________________
> > > >> AMBER mailing list
> > > >> AMBER.ambermd.org
> > > >> http://lists.ambermd.org/mailman/listinfo/amber
> > > >>
> > > >> _______________________________________________
> > > >> AMBER mailing list
> > > >> AMBER.ambermd.org
> > > >> http://lists.ambermd.org/mailman/listinfo/amber
> > > >>
> > > >
> > > >
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> > >
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> > >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Sep 09 2011 - 04:30:02 PDT
Custom Search