[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ccp4bb]: [new subject] TLS refinement



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


> > In pdb are a number of structures, where refmac TLS was used, (converted
> > to Uij s) and these structures show *extraordinarily* low freeR and R for
> > the resolution. So I wonder to what degree your higher rfactors of the CNS
> > model reflect just that, and also, to what degree those low TLS rfactors
> > reflect a truely better model?

As Bart says, it depends what you mean by your model. The mean x,y,z
may or may not have improved, but the estimate of their variations
should have. Furthermore, for medium/low resolutions, the inclusion
of anistropic variations is new, and so comparison with traditional
R/Rfree values (a la DVD's plot) may not be that helpful. Having
said that, if TLS doesn't help you answer your biological question,
then the drop in Rfree is meaningless (except to get your paper
published ;-)

> Furthermore I have just run into a case where the refmac TLS model 
> has lower R and Rfree than full anisotropic refinement.  I found this
> hard to believer, so I replicated the aniso refinement in shelx.
> I'm still scratching my head over this one, but for the moment it seems
> to be real.  The resolution is distinctly margical (1.5A) for full aniso,
> but I would have expected that if the refinement went bad I'd get a 
> lower R but higher Rfree;  that's not what happens.

Certainly, TLS might give you a lower Rfree, but the full aniso should
be able to overfit better. I guess it might be to do with the restraints
on the individual aniso Us. For example, the SPHEricity restraint would
prevent a U going too anisotropic, while a U derived from TLS may
be more anisotropic, which may help to lower R. So you'd need to
remove all U restraints to do a fair comparison (but NOT to generate
results!).

Cheers
Martyn

> (watch for poster at the ACA meeting :-)
> 
> 
> -- 
> Ethan A Merritt       merritt@u.washington.edu
> Biomolecular Structure Center Box 357742
> University of Washington, Seattle, WA 98195
> phone: (206)543-1421
> FAX:   (206)685-7002
> 


-- 


**************************************************************************
*                                                                        *
*                  Dr. Martyn Winn                                       *
*                                                                        *
*      Daresbury Laboratory, Daresbury, Warrington, WA4 4AD, ENGLAND     *
*      Tel: +44 1925 603455    E-mail: m.d.winn@ccp4.ac.uk (personal)    *
*      Fax: +44 1925 603825    E-mail: ccp4@ccp4.ac.uk  (CCP4 problems)  * 
*          URL: http://www.ccp4.ac.uk/martyn/martyn.html                 *
**************************************************************************