It is actually a question of 'adult-crystallographer' addressing a
'baby-coot' ;-)
I also have that problem - and moreover I had it when programming
real space refine in ARP/wARP ages ago.
An elegant solution now is the one implemented by Serge Cohen in my
group, to use a map that has the density of the
part of the model that is not to be refined (and is 'claiming
density') subtracted from the real map.
This way atoms will not go to existing density and can not create a
'steric class'.
This trick - which Alwyn uses since the dawn of crystallographic time
in O rs fit as far as I know -
is computationally quite cheap in our case (we know a priori what to
refine and what not).
In the coot case that map has to be generated every time you refine a
zone, but it will be quick.
A.
PS If Paul wants to try our code (we use clipper anyway)
I am sure your email can motivate him to get in contact ;-)
...he already owes us a visit to get going with some general plans. ;-)
On 29 May 2007, at 3:54, Josiah Obiero wrote:
> Try alternate rotamers first, and then choose the closest to the
> 'right' electron density, before you do 'real space refine'
> Regards,
> Josiah.
>
> Hi,
>
> I use coot almost around the clock. One thing troubles me is that:
> when clicking on 'real space refine zone', coot seems only care about
> the
> 'real space'(Electron density), sometimes it will bring the model to
> the density no matter whether the density was already claimed by
> other
> model atoms or not, resulting in clash and unreasonable geometry. How
> can
> I avoid this?
>
> Sorry for an old crystallographer to ask so babyish question. But your
> help can save me lots of time.
> Thanks
> Yanming
|