[buster-discuss] Refinement of a disordered part of a peptide

[Clemens Vonrhein]

Dear Miguel,

On Wed, May 04, 2016 at 04:02:47PM +0200, Miguel Ortiz Lombard�a wrote:
> I'm refining a structure at ~1.8 A including several copies of a protein
> molecule in the AU.

How many copies are we talking about: 2, 3, ... N?

> For a couple of these the N-terminus is well-behaved and can be
> modelled easily.

Good. Do they (a) show the same conformation, (b) group into a few
conformations or (c) are all different due to different crystal
contacts?

> In the other copies there is some residual, not fully-connected
> density suggesting that the N-termini are mostly in the same
> conformation.

You mean the same conformation within the group of "not
fully-connected density" copies? Or you mean the same conformation as
the well-behaved copies?

> But mostly is not the same than completely and

It is a bit unclear what you mean with that: can you clarify a bit
more?

> trying to model them in the density is not satisfactory: it seems
> clear that they are moving a lot around that conformation.

Do you think they could be multiple conformations or "just" high
mobility?

> I have tried to refine them with lower occupancy after fitting in
> the residual density, but that's not satisfactory either. Leaving
> out these N-termini gives a more correct model in terms of
> stereochemistry but somehow surrenders model completeness and leaves
> behind those blobs.

It seems as if you should try and "harvest" the known conformations
(from the well-behaved copies) via correct use of LSSR NCS-restraints.

> I'd appreciate any ideas that could improve the model in a reasonable,
> not over-fitted way using autoBUSTER.

It depends a bit on

 * Are these termini (1) unstructured, loopy bits that are flopping around
   or (2) a structured (e.g. helix) segment that is connected to the
   rest and moves via some kind of hinge-motion as a (mostly) rigid
   body?

   This could determine if you can use either secondary-structure
   restraints within each segment and/or NCS restraints to other
   copies (for the ririd part) even if they are in a different
   conformation

 * Do you have a single conformation for your well-behaved terminii or
   several distinct groups?

 * How are you using the NCS restraints in BUSTER here: just -autoncs
   (or -autoncs_noprune) or are you defining them by hand?

   You might need to define separate NCS groups for those terminii and
   decide on the most likely pairing/grouping between those. The
   commands are all described at

     http://www.globalphasing.com/buster/manual/gelly/manual/gelly4.html#DEFINE

   but if you could give a detailed description as in

     well-behaved group-1 : A1-34
                            C1-34
                  group-2 : B1-34
                            E1-34

     poorly       group-3 : D1-34  <= similar to group-1
                            F1-34
                  group-4 : G1-34  <= distinct

   we could come up with an example NCS/LSSR description. Or even
   better: if there is a deposited PDB structure that shows similar
   characteristics we could work it into a tutorial.

Sorry for more questions than answers - but maybe it already gives
some ideas.

Cheers

Clemens

-- 

*--------------------------------------------------------------
* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
* Global Phasing Ltd., Sheraton House, Castle Park 
* Cambridge CB3 0AX, UK                   www.globalphasing.com
*--------------------------------------------------------------