[sharp-discuss] Refining Se-met sites

[Clemens Vonrhein]

--z9sQuz+HmDh2hVO4
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Dear Akanksha,

difficult to tell exactly whats going wrong. First I suspect the sites
to be correct from SOLVE. My first attempt would be to actually use
autoSHARP and not SHARP with this data. The information you need to
take care of is

  1. f'/f" values for each dataset/wavelength (hopefully from
     fluorescence scans, but 'standard' Se values are fine too). Do
     _not_ use just the wavelength at the peak and inflection: the
     calculation of these will most definitely give bogus numbers.

  2. a sequence file with the total content of the asymmetric unit
     (attention is only required if you have more than one monomer).

  3. have the SOLVE sites in a *.hatom file (as fractional
     coordinates)

  4. make sure your reflection files (MTZ or SCALEPACK) have the
     correct symmetry (i.e. P212121) and that that symmetry was also
     used in SOLVE to find the sites you put into that *.hatom file.

  5. I would put in the wavelengths in the same order that they were
     collected.

You then start autoSHARP and give it that information. Make sure to
tell it that there are 52 sites (and the *.hatom file with the 30
sites from SOLVE).

The important bits to look out for are:

  0. are there any warnings in the autoSHARP outptu? And what are
     they?

  1. does the automatic interpretation of residual maps keep most of
     the 30 sites and finds additional ones? This would be expected
     ...

  2. does the density modification of the two hands show a significant
     difference in CC?

Especially the last bit will tell you that one of the solutions is at
least on the right track.

This whole procedure will make sure that at least no simple mistake
was done (believe me: very easy, especially if you deal with several
files in different spacegroups with a different number of screw axis
etc).

Let me know what's happening with this approach. If it still doesn't
work we can see how to use data and sites in a different way (in
either SHARP or autoSHARP).

Cheers

Clemens


On Tue, Jan 20, 2004 at 11:47:45AM -0500, gte977x@mail.gatech.edu wrote:
> Dear SHARPusers:
>=20
> I am trying to phase a protein structure using Se MAD data set in P212121=
. I
> found 30 Se sites in SOLVE out of total of 52 with z score =3D 102.56 and=
 fom =3D
> 0.39. The 30 sites were found using SAD data to 3A in SOLVE. I input the =
SOLVE
> output into SHARP and tried refining against the whole MAD data. The refi=
ned
> maps didn't show Se in density for either of the hands.
>=20
> Next, I tried refining sites using SAD data to 2.4 A resolution in SHARP.=
 The Se
> were in density and there were lots of positive peaks in residual maps. I=
 went
> ahead and did solvent optimization. SHARP gave optimal solvent content of=
 38%
> which is not believeable. I expect the solvent content to be around 63% s=
o I
> checked these maps for the secondary sturctures. Map have connectivity bu=
t no
> obvoius sec. str.=20
>=20
> In third set of trials, I tried to use just the remote and peak. Phasing =
stats
> looked much better than SAD but residual maps had lot of negative peaks a=
nd 5 to
> 6 peaks overlapping with HA positions. EDEN maps did have Se in density.=
=20
>=20
> In fourth run,using 2 wav. data (remote and peak) I marked all the f' and=
 f" for
> refinement. In this case there are lot of positive peaks in residual maps=
 but Se
> are not in density again.
>=20
> I am pretty confused by these results. Any suggestions will be greatly ap=
preciated.
>=20
>=20
> Extra info:
> Although I had systematic absences for all three axes I ran solve in P222,
> P21212 and P2221 also. P212121 sats were best.
> =20
> While integrating my data, Cells over the 3 wav.are:
>=20
> Pk =3D 90.3900  163.7399  173.4095   90.0000   90.0000   90.0000
> remote =3D  90.4162  163.2355  173.5318  90.0000   90.0000   90.0000
> inf =3D 90.5415  163.8508  173.8143   90.0000   90.0000   90.0000
>=20
> I reindexed these data sets separately in P212121 and then scaled them us=
ing
> scaleit.
>=20
> Akanksha
>=20
>=20
> _______________________________________________
> sharp-discuss mailing list
> sharp-discuss@globalphasing.com
> http://www.globalphasing.com/mailman/listinfo/sharp-discuss

--=20

***************************************************************
* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
*
*  Global Phasing Ltd.
*  Sheraton House, Castle Park=20
*  Cambridge CB3 0AX, UK
*--------------------------------------------------------------
* BUSTER Development Group      (http://www.globalphasing.com)
***************************************************************

--z9sQuz+HmDh2hVO4
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQFADvX94f+nTPEF25QRAg44AJ92X2HqOZtTbXLZkqdd27gQLuoSKwCcDmoO
+hkelZ+fUUKCgkIP6QrXql0=
=W73C
-----END PGP SIGNATURE-----

--z9sQuz+HmDh2hVO4--