That was the plan initially (sort of). We wanted refmac to generate the
restraints and even do the refinement, but there were problems passing
the information back to Coot. And it was slow to do that.
I am ignorant about Refmac :) But the way the atom selection and
refinement and handled are deeply based on mmdb and clipper code and
extracting/interfacing fortran code would not be easy. It would be a
lot easier to (i) be more clever when looking for links between monomers
and (ii) allow multiple residue ranges to be selected for refinement.
(Groking LINKs from the PDB file will help and Eugene Krissinel has
recently written the code for that.)
P.
On Wed, 2007-05-16 at 09:26 +0800, Charlie Bond wrote:
> Hi Paul,
>
> I'll display the depth of my ignorance here by asking if it is not
> ultimately possible to use the refmac5 (or shelx for that matter) The
> restraints-handling code as a basis for coot, given that they both use
> the same format of information?
>
> Cheers,
> Charlie
>
>
> Paul Emsley wrote:
> > On Tue, 2007-05-15 at 17:06 +0800, Charlie Bond wrote:
> >>> Coot uses the _chem_comp.group to determine the link type.
> >>> It needs to be either L-peptide or D-peptide to be linked as an amino
> >>> acidic chain.
> >> How about non-peptide links, of which my molecule contains a number (15
> >> or so, see PDB 1E9W for an example)?
> >
> > 1E9W: Yikes! I think I'll use that as a torture test for the
> > forthcoming LINK code.
> >
> > I can't help at the moment, sorry. This will be fixed along with
> > glycosylation and branching carbohydrates.
> >
> > Paul.
> >
> > .
> >
>
|