Re: [phenixbb] set_biso_to_wilson_b gone?
Hi Luca,
I've just noticed that in the last version of PHENIX (1.3b-d7) this is not recognised anymore:
refinement.modify_start_model.adp.set_biso_to_wilson_b = true
is this intentional?
I'm afraid, yes. There are some technical reasons for this. You still have at least two ways of doing this: 1) When running phenix.refine : % phenix.refine data.mtz model.pdb modify_start_model.adp.set_b_iso=xxx where xxx is Wilson B values (that you may get from previous phenix.refine runs; phenix.refine reports it). 2) We recently added a tool for PDB file(s) manipulations. Currently it is very simple but will grow based on needs and user's requests. For example, it can read in a model and for selected or all atoms: - modify ADP (randomize, shift, set to a value, scale, convert to/from iso(aniso)tropic); - modify sites (shake, apply rotation / translation); - modify occupancies; - remove selected atoms; - write out a PDB file with modified model. To get started type: phenix.pdbtools -h I'm attaching a copy of a quick manual for phenix.pdbtools (it will be a part of new updated PHENIX documentation). Pavel.
I have a 1.9A dataset that could, in principle, be used to a molecular replacement solution. Phaser finds a solution which yields reasonable map to model agreement, but R factors do not fall down from values greater than 40%. It looked like a P21 (systematic absences are clear), so there was not much options on symmetry. I reprocessed the dataset in P1 and ran phenix.xtriage, and these lines called my attention: --------------------------------------------- Analysing possible twin law : -h,-k,l --------------------------------------------- Results of the H-test on a-centric data: (Only 50.0% of the strongest twin pairs were used) mean |H| : 0.053 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.005 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.447 Estimation of twin fraction via cum. dist. of H: 0.449 --------------------------------------------- Analysing possible twin law : -h,-h+k,-l --------------------------------------------- Results of the H-test on a-centric data: (Only 50.0% of the strongest twin pairs were used) mean |H| : 0.093 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.015 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.407 Estimation of twin fraction via cum. dist. of H: 0.407 --------------------------------------------- Analysing possible twin law : h,h-k,-l --------------------------------------------- Results of the H-test on a-centric data: (Only 50.0% of the strongest twin pairs were used) mean |H| : 0.085 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.013 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.415 Estimation of twin fraction via cum. dist. of H: 0.417 ========= I have never dealt with twinned datasets before. Is there something I could try to do to obtain useful data from this? Lucas P.S.: Unfortunately, growing more crystals is not an option. Alertas do Yahoo! Mail em seu celular. Saiba mais em http://br.mobile.yahoo.com/mailalertas/
Hi Lucas, It would be very informative to see the whole log file, especially the section on intensity statistics. Furthermore, please run the data in P21 through xtriage and find that it is pseudo C222 with a single twin law. I guess you have close to perfect twinning on the operator -h,-h+k,-l (which will look different in P21). Please try phenix.refine twinning=True twin_law=<whatever xtriage gives you> model.pdb data.mtz I hope you have choosen your free set with phenix.refine or iotbx.reflection_file_converter to get twin law compatible free set (done by default in phenix.refine, even if no twin law is specified). Let me know if you have any other problems, Peter Lucas Bleicher wrote:
I have a 1.9A dataset that could, in principle, be used to a molecular replacement solution. Phaser finds a solution which yields reasonable map to model agreement, but R factors do not fall down from values greater than 40%.
It looked like a P21 (systematic absences are clear), so there was not much options on symmetry. I reprocessed the dataset in P1 and ran phenix.xtriage, and these lines called my attention:
--------------------------------------------- Analysing possible twin law : -h,-k,l ---------------------------------------------
Results of the H-test on a-centric data:
(Only 50.0% of the strongest twin pairs were used)
mean |H| : 0.053 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.005 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.447 Estimation of twin fraction via cum. dist. of H: 0.449
--------------------------------------------- Analysing possible twin law : -h,-h+k,-l ---------------------------------------------
Results of the H-test on a-centric data:
(Only 50.0% of the strongest twin pairs were used)
mean |H| : 0.093 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.015 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.407 Estimation of twin fraction via cum. dist. of H: 0.407
--------------------------------------------- Analysing possible twin law : h,h-k,-l ---------------------------------------------
Results of the H-test on a-centric data:
(Only 50.0% of the strongest twin pairs were used)
mean |H| : 0.085 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.013 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.415 Estimation of twin fraction via cum. dist. of H: 0.417
=========
I have never dealt with twinned datasets before. Is there something I could try to do to obtain useful data from this?
Lucas
P.S.: Unfortunately, growing more crystals is not an option.
Alertas do Yahoo! Mail em seu celular. Saiba mais em http://br.mobile.yahoo.com/mailalertas/ _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Hi all, one minor remark:
phenix.refine twinning=True twin_law=<whatever xtriage gives you> model.pdb data.mtz
in the latest phenix.refine it is not necessary to say twinning=True. Providing twin_law=<whatever xtriage gives you> is enough to tell phenix.refine to use twin specific refinement protocols. The keyword "twinning=True" was removed as an element of redundancy. Pavel.
participants (3)
-
Lucas Bleicher -
Pavel Afonine -
Peter Zwart