On Saturday 04 October 2008 18:32:00 Engin Ozkan wrote:
> Hi everyone,
>
> I was in the middle of creating a "Table 1" for a finished structure and
> was puzzled by one number. It is the average B factor, especially in
> the case of TLS-refined structures. In this case, the average reported
> by refmac in the header is the average of the B factors in the pdb file,
> which I assume is the residual B factors (right?) (A quick Baverage
> confirmed the number is the average of numbers in the B column). If the
> reported average is really for residual B factors, isn't this a bit
> misleading? In this case, the average B was 17, but after I created
> total B factors with TLSANL, average B was 25 A^2.
My current standard format for preparing "Table 1" explicitly lists
the residues assigned to each TLS group, and reports B as follows:
Mean B_{iso} + B_{TLS} protein atoms (Å^2): xxx
Mean B_{iso} non-protein atoms: yyy
> I think total B factors should be the one reported, but I am afraid that
> might not have been the practice by many. It is such a petty point, but
> with TLS becoming common, shouldn't users be warned more about such
> issues? Or could Refmac report the average for total B factors?
>
> Engin
>
> P.S. I hate to bring up the controversy about the reporting of
> TLS-refined B factors (residual vs. full anisotropic, or refmac vs
> PHENIX), but it may come to that.
I don't think any controversy is engendered by giving the
formula for the quantity reported in Table 1.
--
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742
