automatically changed refinement parameters
Dear phenixbb, I have some questions regarding refinement parameters: When I run phenix v1.6 via the GUI, starting a fresh refinement session from the main window (I do TLS+isotropic ADP refinement and 5 macrocycles, otherwise everything default), I saw these non-default refinement settings in the run, which I did not specify myself, but the program: adp_restraints { iso { use_u_local_only = True sphere_radius = 1.55 distance_power = 0 average_power = 0 }} In the documentation I found that when doing TLS+isotropic ADP refinement, only Ulocal is restrained, so the first command makes sense. But how is it with the sphere_radius, distance_power and average_power? The default parameters are sphere_radius= 5.0 distance_power= 1.69 average_power= 1.03 What do these parameters mean? What does it mean when these parameters have changed (and why did they change)? And is it ok at all that they changed? And would they have an influence if I modify the start model by setting b_iso to a certain value and then refine "normally" with TLS+isotropic? Another parameter that changed and I did not specify is: Target_weights { wxu_scale = 2.72 } Why did this happen, is it because few cycles before I included optimize_weights? And why would this change the weights permanently? Does it make sense? Thank you for explaining me these things! Sara
Hi Sara,
When I run phenix v1.6 via the GUI, starting a fresh refinement session from the main window (I do TLS+isotropic ADP refinement and 5 macrocycles, otherwise everything default), I saw these non-default refinement settings in the run, which I did not specify myself, but the program:
this is specific to "tls+individual_adp" strategy only. TLS+isotropic ADP refinement protocol is highly optimized by re-refinement of several hundreds structures from PDB to find the optimal refinement parameters. These parameters are coded in phenix.refine and currently cannot be changed. In future versions of PHENIX this will be changes to a user will be able to overwrite these automatically picked values. If you use "optimize_wxu=True" then the value for wxu_scale is adjusted to achieve the best Rfree and Rfree-Rwork. Please let me know if you have any questions! Pavel.
Hi Pavel, thanks for your answer! If there is an "easy" (and for you fast) way to clarify the parameters: Adp_restraints{ iso{ sphere_radius distance_power average_power }} Then I would appreciate some explanation. Sara -----Ursprüngliche Nachricht----- Von: [email protected] [mailto:[email protected]] Im Auftrag von Pavel Afonine Gesendet: Montag, 1. Februar 2010 18:50 An: PHENIX user mailing list Betreff: Re: [phenixbb] automatically changed refinement parameters Hi Sara,
When I run phenix v1.6 via the GUI, starting a fresh refinement session from the main window (I do TLS+isotropic ADP refinement and 5 macrocycles, otherwise everything default), I saw these non-default refinement settings in the run, which I did not specify myself, but the program:
this is specific to "tls+individual_adp" strategy only. TLS+isotropic ADP refinement protocol is highly optimized by re-refinement of several hundreds structures from PDB to find the optimal refinement parameters. These parameters are coded in phenix.refine and currently cannot be changed. In future versions of PHENIX this will be changes to a user will be able to overwrite these automatically picked values. If you use "optimize_wxu=True" then the value for wxu_scale is adjusted to achieve the best Rfree and Rfree-Rwork. Please let me know if you have any questions! Pavel. _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Hi Sara, to answer your question, I'm copying you Ralf's reply to a similar inquiry from a few months ago: **** We use "local sphere restraints". adp_restraints { iso { sphere_radius = 5.0 distance_power = 1.69 average_power = 1.03 } } The basic idea is to restrain each adp to the average of all its neighbors within a sphere of a given radius (sphere_radius = 5). The contribution to the refinement target function is: (u_i - u_j)**2 1 / (r_ij ** distance_power) * ---------------------------------- ((u_i + u_j)/2) ** average_power These terms are computed over a double sum: loop over each atom, loop over all neighbors of the atom. I'm not sure anymore how exactly we arrived at distance_power = 1.69 and average_power = 1.03. You can try different values for distance_power to change the tightness of the restraints as a function of the distance of a pair of atoms. The average_power links the tightness to the absolute value of the adp; average_power=0 turns this feature off. There are some remarks about this in the "ADP refinement" section in this old newsletter article: http://www.phenix-online.org/papers/ccp4_july_2005_afonine.pdf The formulas are a bit different (above is current), but the ideas still apply. In grouped b-factor mode the local sphere restraints are not used. **** Please let me know if you have any questions! Pavel. On 2/2/10 2:13 AM, Sara Züger wrote:
Hi Pavel,
thanks for your answer!
If there is an "easy" (and for you fast) way to clarify the parameters:
Adp_restraints{ iso{ sphere_radius distance_power average_power }}
Then I would appreciate some explanation.
Sara
participants (2)
-
Pavel Afonine -
Sara Züger