Re: [ccp4bb]: arp/warp weard results

[Anastassis Perrakis <perrakis@nki.nl>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


On Wednesday, November 27, 2002, at 07:01 , COUREUX Pierre-Damien wrote:

> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
>
> hello everybody,
> I had some strange results with arp/warp 6 with ccp4i :
> I did one run with the model I was building and arp/warp conserved most 
> of my original coordinates and gave me some new traced chains. That is 
> the good part.
> Then, after rebuilding a little more my model (including 50 more 
> residues for a light chain), I rerun exactly the same job with the same 
> options.
> It looks now that arp/warp has "rebuilt" or at least relocated most of 
> my coordinates at different symmetry positions and my model is broken 
> in pieces.
> Is there something that I missed ?

You did not miss anything. ARP/wARP uses the 'standard' asymmetric unit 
so its 'normal' if it relocates things in an AU definition different 
than 'yours'. Its of course annoying ;-)

As of the molecule being in 'pieces' ARP/wARP 'reshuffle' has a 'dirty' 
algorithm that tries to assemble a 'globular' molecule. If the shape is 
weird that fails. in general though, after side chain fitting it makes a 
new try to assemble the 'real' molecule from chain fragments and 
molecules are usually 'compact'.

A small comment on the general strategy: If ARP/wARP in the first go 
gets more things than you had and at the end of cycling its not 
'stabilized' just run more cycles immediately. If you rebuild yourself 
and then autotrace again that in general will not help much !
The difficult parts are better done manually - ARP/wARP will do the 
'easy' bits and provide a map as good as possible for the more 
disordered / bad regions, but it will not autobuild them usually.

	Tassos