Thanks for the suggestion about using NCS constraint to apply change on one
subunit to other subunits. That should make life much easier.
About the problem of phenix realspace refinement, I figured it out. I used
GUI and when I checked the log file, the "ramachandran_restraints" was
turned off. It is odd since there does not seem to have any buttons to turn
"ramachandran_restraints" off in GUI. Then I tried command line, the
Ramachandron statistics look normal now ("ramachandran_restraints" is on in
log file).
Thanks.
On Fri, May 19, 2017 at 11:01 AM, Pavel Afonine <[email protected]> wrote:
>
>
> Yes, that is what I have been doing. Build one subunit and assemble into
>> tetramer before realspace refinement (with "ncs" constraints). I used
>> tetramer for refinement because I want the distances between the inter
>> subunits interaction partners to be considered. The problem is, whenever I
>> want to make some manual adjustments of the model afterwards, I have to
>> break it into monomer and repeat the whole process again. I assume if the
>> map is in P4 spacegroup, I only need to modify one subunit of this protein
>> and changes will be automatically applied to the other subunits. Am I
>> right?
>>
>
> So you have 4 copies and NCS group consisting of one master (or reference)
> copy and three related by NCS symmetry. Then you make changes in master
> copy and when you run refinement with NCS constraints the changes will be
> propagated onto the related copies by applying NCS constraints. This is how
> this works in phenix.real_space_refine. So no need to make identical
> changes in all 4 copies or go to P4.
>
>
>
>> Another question is, when I tried to refine my model using phenix
>> realspace refinement (energy minimization and adp), the statistics
>> generally become better (map CC, rotamer outlier etc), however, the
>> ramachandron statistics become worse, with increase number of outliers
>> (from below 1% to around 5%) and allowed (from several percent to 10-20%).
>> Do you know what could be the cause? I know I am asking the right person...
>>
>
> This is not expected (about worsening Ramachandran plot outliers). I'm
> sure we can solve this problem if you send me model and map files (and
> resolution) off list (if files too large to send by email please use
> Dropbox or similar file sharing tools).
>
> Pavel
>
>
