VMD-L Mailing List

From: Thomas C. Bishop (bishop_at_tulane.edu)
Date: Tue Mar 31 2009 - 12:15:00 CDT

I had less than a two turns to play with and it was clear that was not
enough... more is better I know that :-)

Cesar Millan seems to have found the answer...
The proper solution appears to be available in charmm

the coor helix command does it.

here's an "unofficial" link to what coor helix does.

http://icbtools.med.cornell.edu/prokink/charmm_helix.html

Tom

On Tuesday 31 March 2009, Peter Freddolino wrote:
> Ah, I see... good point. It makes sense to fit a parametric expression
> to the helix instead... the only problem I can think of with the fitting
> approach is that it might have more trouble with helices that should
> have curvature. Out of curiosity, how long, in your experience, does a
> helix need to be before you start seeing convergence with the calculated
> axis using, say, the principal axes?
>
> Thanks,
> Peter
>
> Thomas C. Bishop wrote:
> > Such an approach is really susceptible to exactly the problem I
> > described. If you chose residues i thru i + N as the helix and calc the
> > principal axes you'll find that as you change N the axis of the helix
> > gradually precesses about what you "know" is the proper axis. each
> > additional residue biases the direction axis direction toward itself.
> > Unless you have several turns the bias is rather strong.
> >
> > The proper method would be to actually fit the mathematical expression
> > for a helix to the Ca atoms of the alpha helix. I don't think this will
> > have the same biasing problem
> >
> > Tom
> >
> > On Tuesday 31 March 2009, Peter Freddolino wrote:
> >> One possibility would be to calculate the principal axes for each helix
> >> (there's convenient code for this at
> >> http://www.ks.uiuc.edu/Research/vmd/script_library/scripts/orient/)
> >> and then calculating the angle using the first principal axis of each
> >> helix.
> >>
> >> Best,
> >> Peter
> >>
> >> Thomas C. Bishop wrote:
> >>> Good question and I'd like to know the answer too.
> >>> based on work some years ago (Bishop and Schulten 1995?)
> >>> I know that unless you have several turns of the alpha helix fitting an
> >>> axis to the helix is very much subject to where you defined the start
> >>> and end fo the helix.
> >>>
> >>> Tom
> >>>
> >>> On Tuesday 31 March 2009, Alison Grinthal wrote:
> >>>> This must be simple but I haven't yet found it (I'm still in the early
> >>>> stages of trying to learn scripting): is there a way to determine the
> >>>> central axis of an alpha helix, and to calculate the dihedral angle
> >>>> between two such axes? If this is explained somewhere or there's a
> >>>> plugin, please direct me. Thanks very much. 

-- 
**********************
Thomas C. Bishop     *
Office: 504-862-3370 *
Fax:    504-862-8392 *
**********************