Which version did you use? I remember the old version of phenix (1.3) will automatically change your setting (wxc and wxu) each round. And v1.5 will automatically choose wxu if you use tls. I haven't tried 1.6 yet. Maybe there are some bugs associated with it? The default wxc is 0.5 and wxu is 1. So the default has tighter weight than your specification. Just use default setting instead.
Nian Huang, Ph.D. UT Southwestern Medical Center On Fri, Mar 12, 2010 at 11:59 AM, Salameh, Mohd A., Ph.D. <[email protected]> wrote: > Yes Im using wxc_scale and wxu_scale but I also use tls refinement,so in > this case if the weights need to be tightly restrained, I should not use > tls refinement! > > ********************************** > Mohd A. Salameh, Ph.D. > Mayo Clinic Cancer Center > Griffin Cancer Research building,Rm 331 > 4500 San Pablo Rd > Jacksonville, FL 32224 > Tel: (904) 953-0046 > Fax: (904) 953-0277 > [email protected] > ********************************** > > -----Original Message----- > From: Phil Jeffrey [mailto:[email protected]] > Sent: Friday, March 12, 2010 12:37 PM > To: Salameh, Mohd A., Ph.D. > Subject: Re: [ccp4bb] phenix target weight refinement > > Use wxc_scale and wxu_scale not wxc and wxu. > If you're doing TLS it will ignore wxu_scale either way. > > > Salameh, Mohd A., Ph.D. wrote: >> *Dear All,* >> >> *In every** refinement** round** the** wxc and wcu** are out of > control, >> although I modify the def file and** change the values** ** of** > wxc=1 >> and wxu=1** so they are tightly restrained**, but it fail**ed** to** >> fix the problem,** wxc and wxu** values**** keep fluctuating,**** in > my >> latest** refinement** round wxc=256!!** wxu=3. The** Stereochemistry >> looks** very** loosely restrained** and the** gap between R-free and >> R-work** is** too big**. I**'**ve been using phenix for almost 4 years > >> and** never had this problem,** or once** I** set the values of wxc > and >> wxu that will be it, never change.** I** also tried to fix wxc from > the >> command line and** didn't** work either (phenix.**refine refine.def >> fix_wxc=1.0).**** I** will certainly** appreciate** any help**. > Thanks, >> Mohd**** ** * >> >> ****************************************** >> >> *****Mohd A. Salameh, Ph.D.* >> >> Mayo Clinic Cancer Center >> >> Griffin Cancer Research building,Rm 331 >> >> 4500 San Pablo Rd >> >> Jacksonville, FL 32224 >> >> Tel: (904) 953-0046 >> >> Fax: (904) 953-0277 >> >> [email protected] >> >> **************************************** >> >
