[phenixbb] Stalled refinement

Yarrow Madrona amadrona at uci.edu
Fri Apr 18 13:59:55 PDT 2014


Thanks Kay,

XDS kicked out a lot of reflections. There were about 23,000 rejected
reflections out of ~ 190,000 collected. I can clearly see another minor
lattice in many frames and I presume that the rejections are coming from
the minor lattice that was not selected. I was thinking of processing with
EVAL15 to see if I get better results. Thanks for your help. Maybe respond
off-line as this is a discussion maybe more suited for another mailing list
(CCP4?).

-Yarrow


On Fri, Apr 18, 2014 at 7:11 AM, Kay Diederichs <
kay.diederichs at uni-konstanz.de> wrote:

> Hi Yarrow,
>
> the problem is that during structure solution, many wrong paths may have
> to be followed until finally identifying the correct path.
>
> So the general answer to this kind of problem is: in some way, your
> parameterization of the experiment is wrong or incomplete.
>
> From what you write, data quality does not seem to be the problem. But:
> did XDS really integrate _all_ the reflections, or only a subset (say,
> every second reflection)?
>
> Check out http://strucbio.biologie.uni-konstanz.de/ccp4wiki/index.
> php/Refinement#what_can_go_wrong_in_refinement.3F
>
> If, after thorough attempts, you fail to find the solution, upload your
> current model, sequence and raw data frames to a Dropbox folder and post
> the link here - there may be people who succeed in processing the data
> nicely, or otherwise can identify the problem based on the data (rather
> than based on your description only).
>
> HTH,
>
> Kay
>
> Am 18.04.14 01:25, schrieb phenixbb-request at phenix-online.org:
>
>> Date: Thu, 17 Apr 2014 16:25:00 -0700
>> From: Yarrow Madrona<amadrona at uci.edu>
>> To: PHENIX user mailing list<phenixbb at phenix-online.org>
>> Subject: [phenixbb] Stalled refinement
>> Message-ID:
>>         <CAMHjG6bPE4q1xWidpE2VwFMJ9qLSqjtLRuLMM1ef9vWBHBfZKg at mail.
>> gmail.com>
>> Content-Type: text/plain; charset="utf-8"
>>
>>
>> Hello,
>>
>> I using the latest stable build of phenx.refine (1.8.4) I recently
>> collected data, processed and obtained an MR solution using phaser. I am
>> stuck trying to refine with an Rfree sitting at 40%
>>
>> I really want to know if the high Rfree is due to poor data quality or if
>> non-crystallographic symmetry involving a near perfect two fold rotation
>> between the two molecules in the ASU could somehow impede refinement.
>> Stats
>> and other information is below. Thank you for any help you can give.
>>
>> -Yarrow
>>
>>
>> Visually, the quality of the data is marginal at best (streaky/ice rings
>> in
>> many frames) despite good processing stats from XDS. Processing with
>> mosflm
>> or HKL2000 managed to index but failed pretty bad in integration and
>> scaling.
>>
>> Phaser gave high TFZ scores for 2 molecules in the asu (see below).
>>
>> Density for a cholesterol like ligand shows up even though not present in
>> the search model.
>>
>> MolRep Self rotation shows rotational symmetry.
>> https://www.dropbox.com/s/2zsajl5o091k50r/CYP142A2-
>> 032814_21_rf%20copy.pdf
>>
>> The 2 molecules in the ASU are related by almost a 2 fold rotation:
>>
>> Rotation matrix for chain A to chain B:
>>
>> new_ncs_group
>> rota_matrix    1.0000    0.0000    0.0000
>> rota_matrix    0.0000    1.0000    0.0000
>> rota_matrix    0.0000    0.0000    1.0000
>> tran_orth     0.0000    0.0000    0.0000
>>
>> center_orth   15.2016    0.5245   33.7070
>>
>> rota_matrix   -0.9860   -0.1636   -0.0309
>> rota_matrix   -0.1659    0.9511    0.2605
>> rota_matrix   -0.0132    0.2620   -0.9650
>> tran_orth      34.3310  -24.0033  107.0457
>>
>> center_orth   15.7607    7.2426   77.7512
>>
>> RMSD, B onto A = 0.0007 after phaser
>> RMSD, B onto A = 0.347 after one round of refinement in phenix
>>
>>
>> Refinement using aniostropically corrected data (ucla web server:
>> Services.mbi.ucla.edu/anisoscale) did not improve the Rfree in
>> refinement.
>>
>>
>> Statistics are listed below:
>>
>> UNIT CELL: 51.487 88.923 89.592 90 97.15 90 P21
>>
>> RESOLUTION     NUMBER OF REFLECTIONS    COMPLETENESS R-FACTOR  R-FACTOR
>> COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
>>     LIMIT     OBSERVED  UNIQUE  POSSIBLE     OF DATA   observed  expected
>>                                     Corr
>>
>>       5.99        8280    1927      2087       92.3%       3.1%      3.3%
>>    8246   35.09      3.5%    99.8*    20*   0.909    1296
>>       4.30       14606    3401      3487       97.5%       3.3%      3.5%
>>   14580   33.37      3.8%    99.9*    11*   0.843    2273
>>       3.53       17961    4244      4445       95.5%       3.8%      3.9%
>>   17944   31.11      4.4%    99.8*    -2    0.789    2721
>>       3.06       21954    5068      5221       97.1%       4.9%      5.1%
>>   21933   24.81      5.6%    99.7*    -2    0.780    3455
>>       2.74       25741    5830      5933       98.3%       7.6%      7.6%
>>   25713   18.88      8.6%    99.5*    -2    0.782    4165
>>       2.51       27859    6311      6483       97.3%      10.8%     10.8%
>>   27824   14.06     12.3%    99.1*    -2    0.774    4385
>>       2.32       31336    6979      7084       98.5%      14.9%     15.3%
>>   31296   10.49     16.8%    98.5*    -4    0.748    5095
>>       2.17       32396    7347      7567       97.1%      22.3%     22.7%
>>   32341    7.46     25.4%    97.3*    -7    0.728    5055
>>       2.05       32254    7339      8047       91.2%      33.1%     33.5%
>>   32075    5.06     37.5%    94.8*    -6    0.724    5155
>>      total      212387   48446     50354       96.2%       7.8%      7.9%
>> 211952   16.57      8.8%    99.7*    -3    0.768   33600
>>
>> Processing with mosflm or HKL2000 managed to index but failed pretty bad
>> in
>> integration and scaling.
>>
>>
>> Phaser:
>>
>> SOLU SET RFZ=27.5 TFZ=24.2 PAK=0 LLG=1711 RF++ TFZ=64.6 PAK=0 LLG=3610
>> LLG=4865
>> -------------- next part --------------
>> An HTML attachment was scrubbed...
>> URL:<http://phenix-online.org/pipermail/phenixbb/
>> attachments/20140417/e09ab4e4/attachment.htm>
>>
>> ------------------------------
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20140418/9d545e3f/attachment.htm>


More information about the phenixbb mailing list