On Monday 08 March 2010 09:28:15 Ethan Merritt wrote:
> I'm trying to come to grips with the recent PDB policy for depositing
> files which have the TLS model expanded out into individual ANISOU
> records. Up until now I've been using tlsextract/tlsanl to prepare
> such files, but now I've just tried using the new refmac option
> TLSO ADDU
> and I'm confused.
Update:
Mitchell Miller pointed out to me that there is another relevant keyword,
one that isn't shown in the GUI: TLSDETAILS WATERS EXCLUDE
Adding that to the input script did indeed cause my waters to remain isotropic.
The result is slightly unexpected, however:
Rfact Rfree FOM -LL rmsBOND rmsANGL
No keywords: 0.2385 0.2884 0.649 272372. 0.0109 1.282
TLSO ADDU 0.2385 0.2884 0.649 272372. 0.0109 1.282
TLSDETAILS WATERS EXCLUDE +
TLSO ADDU 0.2393 0.2903 0.685 272462. 0.0099 1.223
It doesn't surprise me that the R factors are slightly different (0.3% change)
depending on whether the waters are included in the TLS treatment.
But why does the geometry change?
And in particular why such a relatively large change in FOM? (6% change)
Request:
I think the GUI needs a tick box for the include/exclude waters option,
if only so that the user is aware of what's going on.
It is IMHO very bad that nowhere in the PDB file or even the refmac log
file is there any record of _which_ TLS group a water was placed in.
This is a defect in the current PDB ATOM record format!
We really need a field that says "I belong to TLS group #N".
Ethan
> My structure was refined using a pure TLS model.
> That is, all Biso terms for protein atoms were fixed at the Wilson B,
> and then refmac refined the TLS parameters and an overall aniso B correction.
>
> I took my previous "final" refinement run in refmac, added the TLSO ADDU
> keyword to the input script, and reran it.
> So now I have two parallel PDB files:
> the first one produced by refmac/tlsextract/tlsanl (no TLSO option)
> the second one produced by refmac (TLSO ADDU option)
>
> The BISO and ANISOU records in these are not quite the same.
> Not wildly different, but more than I can attribute to round-off error.
>
> Even more confusing, the water molecules in the second file have their own
> ANISOU records. Parvati analysis shows their descriptions are not
> even close to being isotropic. How/what/why did my waters get subject
> to aniso refinement? They are not part of any TLS group, and I didn't
> ask for any individual B refinement of any sort!
>
>
> --
> Ethan A Merritt
> Biomolecular Structure Center
> University of Washington, Seattle 98195-7742
>
--
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742
