[sharp-discuss] question about finding heavy atom position with MR phases

[Clemens Vonrhein]

Hi Sabine,

On Fri, May 16, 2008 at 02:26:17PM +0200, Sabine Schneider wrote:
> Hi everyone,
> 
> I have a question about how to solve the issue with alternative origins 
> when combining SAD and MR phases.
> I got a native dataset to 3.3A with an anomlous scatterer in the ligand. 
> The molecular replacement solution is pretty good but there is different 
> ligand than to the MR model. So I removed the ligand from the model and 
> I am not 100% sure about its orientation at that resolution. Since the 
> ligand contains a heavy atom, I thought that some anomalous data might 
> help. I got a peak dataset with a good looking anomalous signal, to 4.8A 
> (chooped off the data at 4.8A); unfortunately a bit crappy crystals, but 
> I guess it should be suffiecient to determine the HA position.

You could just use the phases from the MR model to calculate an
anomalous Fourier map. This involves something like that

  cad hklin1 peak.mtz hklin2 phaser.mtz hklout cad.mtz <<end_ip
  LABI FILE 1 E1=DANO E2=SIGDANO
  LABI FILE 2 E1=PHI E2=FOM
  end_ip

  fft hklin cad.mtz mapout ano.map <<end_ip
  LABI DANO=DANO SIG1=SIGDANO PHI=PHI W=FOM
  end_ip

  mapmask mapin ano.map mapout ano_asu.map <<end_ip
  XYZL ASU
  end_ip

  peakmax mapin ano_asu.map xyzout ano_asu.pdb <<end_ip
  THRE RMS 5.0
  NUMP 8000
  end_ip

(obviously with correct file and column names). You could restrict the
resolution to using only low-res data (say 20-5.0).

You don't need to chopp off the data at 4.8A just because the
anomalous goes only to 4.8A: I would use your normal/favourite cut-off
criteria (I/sigI, completeness ... whatever) to decide where to cut
the data. The 4.8A cut-off is only important for the HA detection step
- but if you use an anomalous Fourier map as above you won't need to
search for sites anyway.

Once you have the sites, put them into a *.hatom file in your
sharpfiles/datafiles directory - in fractional coordinates:

  coordconv xyzin ano_asu.pdb xyzout ano_asu.frc <<end_ip
  INPU PDB
  OUTP FRAC
  end_ip

Then in autoSHARP you can select that file (everything else stays
identical) and it will skip the HA detection step (using the given
sites as starting values).

> I used ShelxD to find the sites (2 HA per asu) and the solution looks 
> somewhat sensible and I also get a nice peak in the difference patterson 
> (better than in the anomalous patterson).  The crystals are tetragonal 
> and I therefore have four alternative origins.

Plus the enantiomorph of course.

> So my first question is, to run SHARP using the HL coefficients from
> phaser, what initial site do I put in or how to I make sure that the
> sites I got from the difference patterson / ShelxD are on the same
> origin as the MR phases? (I scaled the two datasets together with
> scaleit and than run phaser on the native data )
> 
> I put in 0 0 0 as initial site in SHARP

Correct.

> and it comes up with two sensible looking sites in the residual
> maps. I fed the sites found in the residual maps back into SHARP,
> but how come that I still have to shift my MR model z = 0.5 that it
> is matching the electron density calculated from eden.mtz?

Is it possible that your MTZ file contains e.g. the spacegroup P41
(which is indistinguishable from P43), but phaser detected the correct
spacegroup P43? This could be the reason for that Z=0.5 shift ...

Make sure that SHARP runs with the correct spacegroup - the easiest is
to change your MTZ file to the spacegroup you got as correct solution
in PHASER, e.g. with

 cad hklin1 your.mtz hklout your_P43.mtz <<end_ip
 LABI FILE 1 ALL
 SYMM P43
 end_ip

> So I shifted the sites z= -0.5 and repeated the SHARP run. It comes
> up again with the 2 more sites in the residual maps, but they are
> just the same again on a different origin? And to match my MR model
> to the ED I have to shift the origin of my MR model again? I thought
> that if I use the phases from the MR for the calculation of the
> residual maps and get some sites from there, they should be on the
> same origin than the model? Moreover the sites are not making sence
> in terms of the location in the model so I guess they are wrong
> than? I have to protein-ligand complexes in the asu with one HA each
> and the distance between the active site of the proteins in the asu
> and between the two sites found by SHARP are similar enough?

Yes - sounds a bit confusing.

> Would be very much appreciated, if someone could tell me, where I am 
> going wrong!

I would check the screw axes in your reflection file(s) and from the
MR solution. I would also use an anomalous Fourier for finding the
sites as well - to double check the sites you get from SHARP+HL(MR)
(which is a good idea nevertheless!).

Cheers

Clemens

> 
> Thanks a lot for your help in advance!
> 
> Sabine
> 
> _______________________________________________
> sharp-discuss mailing list
> sharp-discuss at globalphasing.com
> https://www.globalphasing.com/mailman/listinfo/sharp-discuss
> 

-- 

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