*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
On Friday 15 December 2006 10:00 am, Miller, Mitchell D. wrote:
> I was surprised to hear that others have had so much trouble with
> the deposition of phases at the RCSB site. At JCSG, we have been
> depositing experimental phases since Dec 2003/Jan 2004 without
> any problems from the RCSB.
I think part of the issue is the format in which you submit your
structure factors. The RCSB submission process leads one to believe
that you can submit *.mtz files, but this is very misleading.
We were dismayed to learn that many of our structure factor depositions
had been stripped of Fcalc, Iobs, and other fields present in the
original *.mtz files. The RCSB script apparently looks for specific
column labels rather than simply archiving everything in the file.
We brought this issue to their attention, but then switched over to
depositing cif files. So I cannot say for sure whether this is
still a problem or not. Safest not to deposit *.mtz files, I think.
On the other hand, we have been able to deposit arbitrary reflection
data in cif files, modulo an acceptable level of misunderstanding
or mis-parsing of the contents.
Ethan
> Our standard for MAD/SAD structures is to have several data
> sections that are defined in the structure factor file.
> 1- loop over merged refinement data / results Fobs, SigFobs, Fcalc,
> PhiCalc, status 2- loop over scaled but unmerged intensities for the
> refinement wavelength 3- loop over scaled but unmerged intensities
> for 2nd wavelength 4- continue with similar loops over all
> wavelengths
> .....
> 5- loop over HL-coefs for the experimental phases
> 6- loop over DM modified phases (before including partial
> models from iterative building) phase, fom and
> _refln.pdbx_fom_weighted_fmap The last item is a pre-weighted
> amplitude that can be used with the DM phases to calculate the MAP
> that was used for the start of auto-building.
>
> In cases were we have MAD phases from one crystal and our final
> refinement is against a second higher resolution crystal, we include
> the scaled, unmerged original index intensity from both crystals /
> all wavelengths. We use the refln.crystal_id value to identify which
> data comes from which crystal in the file.
>
> Regards,
> Mitch
>
> P.S. for those that are interested in the mmCIF format details
> for our structure factor files, the file structure starts with
> data_rxxxsf for the refinement data (where xxxx is the PDB ID).
> Then subsequent data values of the form data_rxxxxAsf,
> data_rxxxxBsf, etc... By having them in separate data_ section
> we can have slightly different unit cells for the different
> wavelengths / crystals.
>
> Here are a few examples....
> 2AML is a standard 2-wavelength MAD case
> http://www.pdb.org/pdb/explore/explore.do?structureId=2AML
>
> and 2HUJ is an example of a case with 2 crystals. One for
> 3-wavelength MAD and the second for refinement.
> http://www.pdb.org/pdb/explore/explore.do?structureId=2HUJ
>
>
>
>
> ________________________________
>
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf
> Of Erik Vogan Sent: Friday, December 15, 2006 7:34 AM
> To: William Scott
> Cc: [email protected]
> Subject: Re: [ccp4bb]: PDB and phases ...
>
>
> Bill (and list),
>
> When I initially deposited the diffraction data for structure I
> solved as a graduate student, I met with the same "they won't take
> them" attitude regarding my experimental phasing data (full MAD/MIR
> data sets, not phases). It was after some encouragement from people
> at the 2004 Diffraction Methods GRC that I decided to take the
> attitude "they bloody will take them." After all, the PDB works for
> _us_, not the other way around.
>
> In any case, after writing a couple of very strongly worded
> emails, the RCSB did, in fact, accept my experimental phasing data
> sets.
>
> Although this does not in any way specifically address your
> attempt to contribute phases, the mmCIF format contains both
> _refln.phase_meas and _refln.phase_calc items, so there should be no
> reason for the fine folk at the PDB to refuse your (very reasonable)
> request that they accept phases along with the amplitudes.
>
> Finally, for what it is worth, the EBI seemed much more willing
> to accept heavy atom data sets. They did not even raise a tiny fuss
> about the issue. However, if you have already contributed a
> structure to a specific site (say, RCSB for instance), then you will
> need to take the issue up with that site, as they are the only ones
> who can update the initial deposition.
>
>
> Erik Vogan, Ph.D.
> Structural Biology and Computational Chemistry
> Wyeth Research
> 200 Cambridgepark Drive, T6011E
> Cambridge, MA 02140
> (617) 665-5501
>
> >>> "William Scott" <[EMAIL PROTECTED]> 12/15/2006 10:05:00
> >>> AM >>>
>
> *** For details on how to be removed from this list visit the ***
> *** CCP4 home page http://www.ccp4.ac.uk
> <http://www.ccp4.ac.uk > ***
>
>
> I've tried to give them phases. They won't take them. I think it
> would be really helpful to have the experimental map (or a
> composite-omit map in the case of MR). I don't think there is a way
> of representing the latter as phases, so the option to submit phases
> and/or map should be available.
>
> Peter Adrian Meyer wrote:
> > *** For details on how to be removed from this list visit the ***
> > *** CCP4 home page http://www.ccp4.ac.uk
> > <http://www.ccp4.ac.uk > ***
> >
> >> I agree that more information to referees is desirable, & I've
> >> often
> >
> > wanted to see the coordinates and the structure factors to check
> > things. But to follow up after publication, there are still a lot
> > of structures deposited without structure factors - why is
> > deposition of structure factors not compulsory? And who can change
> > things so that
> >
> > This is probably an idiot newbie question, but the vast majority of
> > structures I've seen in the PDB w\ "structure factors" only have
> > amplitudes. Granted, you could calculate phases from the model;
> > but shouldn't a structure factor be a complex number instead of a
> > real number?
> >
> > I'd think there should be the experimental phases, or at least
> > model phases with the weighting scheme the authors used for their
> > final maps for it to be called a structure factor...but evidentally
> > I'm wrong.
> >
> >
> > Pete
> >
> >
> > Pete Meyer
> > Fu Lab
> > BMCB grad student
> > Cornell University
--
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle WA
