[phenixbb] Average B-factors when using TLS

Pavel Afonine pafonine at lbl.gov
Wed Oct 13 14:41:25 PDT 2010

  Hi Phil,

I'm replying since I implemented TLS refinement in phenix.refine. 
Honestly, I've never done such comparisons, so I can't really comment on 
this thoroughly. The ADP refinement in phenix.refine is described here:

starting at page 23 (this is the very latest source), and here (older)


and here (oldest):


and for completeness:


I'm happy to further discuss it, see your suggestions and do tests if it 
comes to it (although I'm in travel now, and I will return back from UK 
and France sometime around Octber 20, so I may not be interactively 
responsive during this time). I'm interested in feedback and 
constructive suggestions.

Thanks for questions and all the best!

On 10/13/10 11:20 AM, Phil Jeffrey wrote:
> I'm having a discussion with a colleague here about what we expect to 
> see on a model with and without TLS being used.  I had the expectation 
> that if you'd compared both models the Biso values reported in the 
> ATOM line for the same atom w/ and w/o TLS would be somewhat 
> comparable.  Not the same, because the ADP model is different, but 
> similar.
> However in the case of the TLS-refined model the average B-factors are 
> systematically higher over the whole structure (86 A^2 vs 69A^2 for 
> one polypeptide chain).  Is this reasonable ?  Expected ?
> Phil Jeffrey
> Princeton
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

More information about the phenixbb mailing list