Saturday, 31 December 2011

sugar and coot

From: Jan van Agthoven
Date: 22 November 2011 00:16


Hi everyone!
Does anyone know if there is a way of auto-refining a sugar in Coot?
Jan

----------
From: Dirk


Hi,

I'm not sure if that is what you are looking for, but if you want to use the "real space refine zone" function in coot with sugars that is possible.
For a monosaccharide you will need a cif dictionary with the restraint definitions - in some cases present in the refmac monomer database, but be careful not all cif files there are correct. If none is present or if your monomer is disrupted during refinement you can look at the small molecule databases (Hic-Up etc) or calculate one with PRODRG  (http://davapc1.bioch.dundee.ac.uk/prodrg/). Load it with "File" "Import CIF dictionary and real space refinement should work.
If you try to refine a polysaccharide it's a bit more difficult since coot will ignore any glycosidic bond restraints - if two sugars are close by and one has well defined density and the other one not it might even place them on top of each other.
I like to use the anchor tool on the bonding atoms in those cases.
The other option for a polysaccharide is to use PRODRG to generate a single "monomer" out of your polysaccharide.

In any case at the end you have to inspect the geometry manually since the restraint definitions are often incomplete.

Greetings

Dirk




----------
From: Joel Tyndall


You can also try Jligand to generate your cif file

J

 


----------
From: Paul Emsley


Such a tool (if I understand correctly what you want) does not yet exist. However, I believe that things are not in as poor a state as one might infer from other posts.

Using the pre-release, you can

1) ideally position a NAG onto ASN
2) use multi-residue torsion to position a NAG by rotation of ASN chi angles
3) use the Ligand builder to build new carbohydrate monomers (it is my experience that PRODRG respects the wedge bonds - but I have heard that this is not always the case)

Using latest stable or later you can

4) multi-residue refinement (or sphere refine), which does a link analysis based on the coordinates of the residue selection (note that bonding atoms have to be within 3A to be considered a real glycosidic linkage)
5) add addition pseudo plane restraints to encourage ideal chair (in my experience, at low resolution or with poor density, carbohydrates tend to wander off and twist up without these)

The JLigand-Coot interface has stalled a bit, but I hope should be available for the Study Weekend.

The Refmac-Coot link-up using LINKR is under discussion with Garib and Eugene.

I would recommend against making a synthetic disaccharide carbohydrate with PRODRG.

Regards,

Paul.

----------
From: Heping Zheng


The issue I would raise here is not just for coot, but also for PDB format
itself. Last time I asked on ACA2011, Dr. Berman said polysacchride
residues in PDB is still not yet uniformly represented.

The issue I have is to polysacchride chains refinement in general, but
here more specific about branch chain of polysacchride chains. For
example, if you think about the case of N-glycosylation
Asn-NAG-beta1,4-NAG-beta1,4-Man-(alpha1,3-Man)-alpha1,6-Man ...

Now you have a branch chain, Now if I assume the following resseq
ASN A135
NAG A500
NAG A501
MAN A502
MAN A503
MAN A504

In my coot, version 0.6 revision 2540, You will be able to refine
A500-A503 as a whole since they are consecutive in resseq. But when I need
to refine NAG A500 together with ASN A135 locally, simply specify the
range is not enough, I need to specify a small radius to get it work, even
though the covalent link is defined in LINKR record. The same applies when
I want to refine just A502-A504 alpha1,6 glycosidic bond without refine
A503.

I am just wondering what is the state of the art for branch chain handling
in coot or another interactive refinement program.

best regards,
Heping Zheng

----------
From: Paul Emsley


Stuck email... try again...
 indeed. Quite so.  As I said (or more accurately, implied) LINKR recognition is
an on-going issue (i.e. it doesn't work yet). Yes. Well, I don't know much about about other programs, but in coot you can
use refine-residues or Sphere Refinement.
Even in 0.6.

Paul.


No comments:

Post a Comment