On Friday 27 August 2010, Frank von Delft wrote:
> The requirement sounds extremely suspect: every atom in the structure
> contributes to every reflection, so refining "only some atoms" makes as
> little mathematical sense as refining against "only a subset of
> reflections".
I agree with you that the requirement sounds dubious.
But the specific argument you make is not quite right.
Two common counter-examples are real-space refinement and rigid-body
placement of a known fragment relative to an existing partial model.
> Especially not Bfactors, which are after all where errors get pushed
> into. So if you refine Bfactors for only some atoms, all you're doing
> is making the refinement program shove all the errors in the *whole*
> structure into the Bfactors of a few atoms. You'd have to explain
> extremely carefully why those atoms deserve that -- and even then I
> doubt anybody would believe you.
Indeed.
Ethan
>
> phx.
>
>
>
> On 26/08/2010 22:19, Hailiang Zhang wrote:
> > Thanks a lot Ethan, I will give it a try.
> >
> > Best Regards, Hailiang
> >
> >> On Thursday 26 August 2010 11:56:39 am Hailiang Zhang wrote:
> >>
> >>> Hi,
> >>>
> >>> I want to refine B factors for several residues only (all the other B
> >>> factors and all coordinates fixed, I know it sounds weird but there is a
> >>> reason to try that).
> >>>
> >> Maybe you could tell us what this reason is?
> >>
> >>
> >>> Is there anyway CCP4 can do this? Thanks for any suggestions!
> >>>
> >> Suggestion 1)
> >>
> >> Calculate structure factors for the entire rest of the model.
> >> Include these as F_partial contributing to the refinement of a model
> >> containing only your residues of interest. In this refinement,
> >> refine only the B terms.
> >>
> >> Caveat: I think you will encounter problems with how to handle
> >> the bulk solvent correction. Perhaps that must be included in
> >> F_partial also, and omitted from the subsequence mini-refinement.
> >>
> >> Suggestion 2)
> >>
> >> - Place your residues of interest in a single TLS group.
> >> - Do not assign any other atoms or residues to a TLS group.
> >> - Refine the entire model using refmac in TLS refinement mode.
> >> Choose 5 or 10 cycles of TLS refinement, but 0 cycles of
> >> coordinate/Biso refinement. Disregard all output other than the
> >> refined TLS description for the B factors in your residues of
> >> interest.
> >> - Use TLSANL to expand the TLS parameters back to individual
> >> Biso if you like.
> >>
> >>
> >>
> >>
> >>> Best Regards, Hailiang
> >>>
> >>>
> >> --
> >> Ethan A Merritt
> >> Biomolecular Structure Center, K-428 Health Sciences Bldg
> >> University of Washington, Seattle 98195-7742
> >>
> >>
> >>
>
