<BODY contentEditable=true onload=parent.init()><P>Weerd, </P>
<P>my source code shows that the default is ml:</P>
<P>cd /net/rosie/scratch1/phzwart/CCTBX/Sources/mmtbx/mmtbx/bulk_solvent/</P>
<P>more bulk_solvent_and_scaling.py</P>
<P>master_params = iotbx.phil.parse("""\<BR> bulk_solvent = True<BR> .type = bool<BR> anisotropic_scaling = True<BR> .type = bool<BR> statistical_solvent = False<BR> .type = bool<BR> k_sol_b_sol_grid_search = True<BR> .type = bool<BR> minimization_k_sol_b_sol = True<BR> .type = bool<BR> minimization_b_cart = True<BR> .type = bool<BR> target = ls_wunit_k1 *ml<BR> .type = choice</P>
<P>Pavel developed some heuristics governing optimal settings in certain cases which might come into play here. I suspect it might be due to model size, resolution or other factors. If not, I am puzzled.</P>
<P>Cheers</P>
<P>Peter </P>
<P><BR><BR>----- Original Message ----- <BR>From: Jianghai Zhu <jzhu@cbr.med.harvard.edu> <BR>Date: Sunday, March 25, 2007 5:18 pm <BR>Subject: Re: [phenixbb] ls_wunit_k1 <BR>To: PHENIX user mailing list <phenixbb@phenix-online.org> <BR><BR>> Hi Peter, <BR>> <BR>> You are right. Somehow, the default target function in <BR>> refinement.bulk_solvent_and_scale is LS, not ML. <BR>> <BR>> Jianghai <BR>> <BR>> +++++++++++++++++++++++++++++++ <BR>> Jianghai Zhu, Ph.D <BR>> CBR Institute for Biomedical Research <BR>> Department of Pathology <BR>> Harvard Medical School <BR>> 200 Longwood Ave., Boston, MA 02115 <BR>> Ph: 617-278-3211 <BR>> Fx: 618-278-3232 <BR>> +++++++++++++++++++++++++++++++ <BR>> <BR>> <BR>> <BR>> On Mar 25, 2007, at 8:18 PM, Petrus H Zwart wrote: <BR>> <BR>> > Hi Jianghai, <BR>> > <BR>> > Interesting, I though that by default an ML target is used. <BR>
> > Something must have happened somewhere. <BR>> > <BR>> > Check the parameter file and look for the section on bulk <BR>> solvent <BR>> > scaling ( refinement.bulk_solvent_and_scale ). A target function <BR>> <BR>> > can be defined, you have the choice between ml and ls_wunit_k1, <BR>> ml <BR>> > is (I think) selected by default. <BR>> > <BR>> > There is also a switch present that triggers the refinement of <BR>> bulk <BR>> > solvent parameters using a least squares target if the r value <BR>> is <BR>> > larger than a certain value: <BR>> > <BR>> > ls_bulk_solvent_and_scale_if_r_greater_then = 0.99 <BR>> > <BR>> > Changing the parameter you propose changes the least squares <BR>> target <BR>> > type if the main target selected is ls and has (as far as I <BR>
> > understand) no influence on bulk solvent scaling. If you want, <BR>> you <BR>> > can send a log file for a closer look. <BR>> > <BR>> > HTH <BR>> > <BR>> > Peter <BR>> > <BR>> > <BR>> > <BR>> > <BR>> > <BR>> > <BR>> > ----- Original Message ----- <BR>> > From: Jianghai Zhu <jzhu@cbr.med.harvard.edu> <BR>> > Date: Sunday, March 25, 2007 2:58 pm <BR>> > Subject: [phenixbb] ls_wunit_k1 <BR>> > To: PHENIX user mailing list <phenixbb@phenix-online.org> <BR>> > <BR>> > > Hi, <BR>> > > <BR>> > > I happen to notice that the target function of "Bulk solvent <BR>> > > correction and scaling" in a phenix.refine run is ls_wunit_k1. <BR>> > > Does <BR>> > > that mean LS function for bulk solvent correction and scaling is <BR>
> > > the > default in phenix.refine? If I uncheck ls_wunit_k1 in <BR>> > > refinement.ls_target_names in my parameter file, will <BR>> > > phenix.refine <BR>> > > use the ML function or something bad will happen? <BR>> > <BR>> > > <BR>> > > Jianghai <BR>> > > <BR>> > > +++++++++++++++++++++++++++++++ <BR>> > > Jianghai Zhu, Ph.D <BR>> > > CBR Institute for Biomedical Research <BR>> > > Department of Pathology <BR>> > > Harvard Medical School <BR>> > > 200 Longwood Ave., Boston, MA 02115 <BR>> > > Ph: 617-278-3211 <BR>> > > Fx: 618-278-3232 <BR>> > > +++++++++++++++++++++++++++++++ <BR>> > > <BR>> > > <BR>> > > <BR>> > > <BR>> > <BR>> > _______________________________________________ <BR>> > phenixbb mailing list <BR>> > phenixbb@phenix-online.org <BR>
> > http://www.phenix-online.org/mailman/listinfo/phenixbb <BR>> <BR>> </P></BODY>