From jpaint@u.washington.edu Wed Aug 31 14:28:22 2005
Date: Wed, 24 Aug 2005 13:25:40 -0700
From: Jay Painter <jpaint@u.washington.edu>
To: ccp4bb@ccp4.ac.uk
Subject: Re: [ccp4bb]: problems with  - refmac5

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


You can submit a isotropic refinement (no TLS groups) to the TLSMD
server at:

http://skuld.bmsc.washington.edu/~tlsmd/index.html

It will partition your structure into multiple TLS groups for
refinement.  After your structure has been processed by the server, you
may select the number of TLS partitions you want to use and generate
input files for REFMAC5 TLS refinement.


On Wed, 2005-08-24 at 13:04 -0700, Alex Bankovich wrote:
> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
> 
> 
> Mark,
>          I had a similar problem with a poorly ordered domain in my 
> structure.  I found that leaving that domain out of TLS refinement 
> allowed for better refinement as evidenced by a decrease in 
> Rfree.  Since you have high resolution, you might benefit from 
> breaking your structures up into even smaller TLS groups.  I used CNS 
> refined B values as a basic guide, but you can try different 
> criteria, whether it be specific domains, strands and helices, or 
> removal of loops from refinement.  Hope that this helps.
>                                          Alex
> 
> At 12:12 PM 8/24/2005, you wrote:
> >***  For details on how to be removed from this list visit the  ***
> >***          CCP4 home page http://www.ccp4.ac.uk         ***
> >
> >
> >Refmac5 TLS refinement question.
> >
> >I'm trying to refine a 250 residue protein with a novel ligand 
> >complex using geometry restraints, individual B, with overall matrix 
> >weight set to 0.1 . One domain is very well ordered, the 2nd less 
> >well. I''ve defined the two domains as separate TLS groups. My data 
> >goes to 1.7 A. The structure was solved by MR using PHASER. The 
> >model is essentially complete now.
> >
> >When refining with TLS the log file gives a warning that there is  a 
> >problem making U factors positive (they are around -1 or smaller). I 
> >understand that a negative U is physically impossible, but I cannot 
> >figure out what's wrong with my model/refinement setup.
> >
> >I'd greatly appreciate advice to solve this problem.
> >
> >Thanks in advance,
> >mark,
> >mayerm@mail.nih.gov
> >
> >--1234--
> 
>