[phenixbb] ncs restraint
Engin Ozkan
eozkan at stanford.edu
Mon May 25 16:26:41 PDT 2009
Hi Pavel,
It is very rarely sensible to disagree with automation, definitely in
this case. My problem was that the automated NCS was including bits
that were clearly violators of NCS (This was with 1.3 final release). I
will be looking forward to any improvements in this respect. Here are a
couple questions/suggestions:
1. Does automated NCS picking only work based on the input pdb? Should
it not consider electron density maps as well for true automation (which
may not be easy to implement)? Otherwise, regions that should eventually
be excluded from NCS will be initially restrained, and stay that way
cycle after cycle by the automated NCS picking, until some person
intervenes and remodels the pdb. Automated analysis of model and
electron density for violations of NCS would be a god-send for those
large structures with many, many (> 10) NCS chains. (How about a
phenix.analyse_complex_ncs instead of phenix.simple_ncs_from_pdb :))
2. Any automation on restraint weights (possibly based on Rfree
behaviour, or something even more complicated)? That would be nice. And
definitely not excluding NCS restraints on ADPs (It seems to me
commonplace to have one or more NCS related groups to be more
"flexible", more "fuzzy". It helps to relax NCS-ADP restraints on those
chains).
Thanks,
Engin
On 5/25/09 3:41 PM, Pavel Afonine wrote:
> Hi Engin,
>
> since the model gets improved during refinement (hopefully), it is a
> good idea to update NCS restraints based on improved model, so the
> default behavior.
>
> The use of NCS restraints in phenix.refine is explained in the manual:
> http://www.phenix-online.org/documentation/refinement.htm#anch24
>
> I admit that the automatic NCS determination has a room for improvement,
> but once again: ideally it should be done fully automatically and be
> updated as model gets improved. It is in my list...
>
> As it is now it's always good to check the automatic choice (as Ralf
> mentioned).
>
> Pavel.
>
>
> On 5/18/09 3:33 PM, Engin Ozkan wrote:
>
>> Also, don't forget to add ncs.find_automatically=false
>>
>>
>> I have run many runs with NCS defined with proper syntax, just to be
>> overwritten by NCS derived automatically by phenix, until I realized
>> what was happening. I have always thought that if I define the NCS
>> restraints, phenix should follow it, not try to "improve" upon it. What
>> is the reason for this default behavior?
>>
>> Engin
>>
>>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb
>
More information about the phenixbb
mailing list