[phenixbb] MAD merged dataset deposition

Mitchell D. Miller Mitchell.D.Miller at rice.edu
Sun Mar 3 06:33:51 PST 2019


Hi Eugene,
   You can deposit more than one data loop for your MAD deposit in
your mmCIF formatted structure factor file.

We used to do this regularly at JCSG. Our mmCIF formatted
reflection file would contain multiple data_XXXX loops to
contain:
1) loop over refinement data (f, sigF, FC, PhiC, and now you could
add your final refinement map coefficients as well e.g.
_refln.pdbx_FWT
_refln.pdbx_PHWT
_refln.pdbx_DELFWT
_refln.pdbx_DELPHWT
2) for each wavelength (and sometimes for each crystal when different crystals
used for phasing and refinement) a separate loop containing I and sigI  
of the scaled
but unmerged original index data. This would allow someone to later go back
and recompute your phasing run and would also allow testing different
space group possibilities.
3) Next would be a loop over the phasing output (e.g. HL-coefs)
4) The final loop would be over the density modified experimental map
prior to autobuilding.

See as an example 4sfv for a 3 wavelength MAD case with 1 crystal and
4qhz as a 2 wavelength MAD case with data from 2 crystals.

We did this all with custom data extractions scripts from our database,
but the new web-based PDB-extract tool makes it pretty easy to prepare
a multiple data_loop formatted mmCIF file for deposition.
You may still need to do some hand editing/ combining if you
want to include the phasing results and density modification output.


Regards,
Mitch


Quoting Eugene Osipov <e.m.osipov at gmail.com>:

> Hello, everyone,
> I have solved the structure of heme containing protein using anomalous
> signal from Fe atom (~1.739 anstroms) and further refined it using native
> dataset (0.9 angstroms).
> Structure was solved in AutoSol. The question is "What is the best way to
> deposit mtz with both datasets?"
> Should I merge and scale them simultaneously or can I deposit output file
> from AutoSol overall_best_refine_data.mtz, which contains F,FP, Rfree and
> HL coefficients?
> P.S.: please, respond in PM, I have only digest subscription on phenixBB.
>
> Thank you in advance!
> --
> Evgenii Osipov
> Laboratory for Biocrystallography,
> Department of Pharmaceutical Sciences,
> KU Leuven O&N2




More information about the phenixbb mailing list