AMBER Archive (2007)

Subject: Re: AMBER: Specific/Non-specific dihedral issue in Amber

From: Narayanan Veeraraghavan (narayanan.veeraraghavan_at_gmail.com)
Date: Tue Jul 17 2007 - 17:01:10 CDT


I'm sorry to drag this issue, but I still don't see the bottomline.

You said:
> Same idea for specific vs. non-specific: if you put in specific
> parameters for a torsion, you should include specific versions for all
> multiplicites (even those that might have a zero force constant). Basically,
> don't leave LEaP any choice about which set (general, specific, frcmod) of
> parameters to use.

The question remains, if there exists a specific dihedral with say
multiplicity 1 and a generic dihedral with multiplicity 3, for e.g.
H1-CT-CT-OS 1 0.25 180 1
X -CT-CT-X 9 1.40 180 3

should we include the generic dihedral with a multiplicity say 3, with
the force constant as given in the generic dihedral definition, eg:
H1-CT-CT-OS 9 1.40 180 3

or with a zero force constant (which basically implies disregard the generic)?
H1-CT-CT-OS 1 0.00 180 3

Could you tell what *ought* to be the rule?

Thanks,
-NV
BTW, the system is a nucleic acid (sugar) with parm99.dat ff.

On 7/16/07, David A. Case <case_at_scripps.edu> wrote:
> On Mon, Jul 16, 2007, Narayanan Veeraraghavan wrote:
>
> > I once again looked into this issue of specific and generic dihedrals.
> > It seems that in some instances tLeap choses to include both generic
> > and specific dihedrals (with different multiplicities) whereas in
> > other instances it simply ignores the generic dihedrals (even when the
> > multiplicities are different).
> >
> > For example:
> > H2'1 - C2' - C3' - O3' 0.250 0 1 (H1-CT-CT-OS)
> > H3' - C3' - C2' - O2' 0.250 0 1 (H1-CT-CT-OH)
> >
> > The above, have just the specific dihedral entry in the topology, but
> > not a non-specific inspite of the fact that the generic dihedral
> > X-CT-CT-X has a multiplicity of 3.
> >
> > Cases where AMber includes non-specifics
> > H4' - C4' - C3' - O3' 0.156 0 3 (H1-CT-CT-OS - non specific)
> > H4' - C4' - C3' - O3' 0.25 0 1 (H1-CT-CT-OS - specific)
> >
> > H5'1 - C5' - C4' - O4' 0.156 0 3 (H1-CT-CT-OS - non specific)
> > H5'1 - C5' - C4'_O4' 0.25 0 1 (H1-CT-CT-OS - specific)
> >
> > Could someone kindly tell me if I am missing something?
>
> This is a known bug of LEaP: basically, the "rule" is: if you modify
> (in a frcmod file) one of the multiplicities in a torsion, you need to modify
> all of them. Same idea for specific vs. non-specific: if you put in specific
> parameters for a torsion, you should include specific versions for all
> multiplicites (even those that might have a zero force constant). Basically,
> don't leave LEaP any choice about which set (general, specific, frcmod) of
> parameters to use.
>
> I think the peptide/protein force fields have been examined for their
> compliance with this "rule", but the same might well not be true for sugars.
> I've lost track of which force fields you are using, (or even what kind of
> molecule.) For now, the "fix" is to make the parameter files more complete,
> so that the sort of ambiguity referred to above does not enter.
>
> As we transition from sleap to tleap, Wei should try to see if we can get
> sleap to follow well-defined rules for this sort of a situation. Some one
> could also earn gold-stars by taking the time to really figure out what tleap
> does and why, but I'm not holding my breath.
>
> ...regards...dac
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

-- 
n_at_r@y_at_n@n
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu