Wednesday, 29 February 2012

MTZ file

From: Uma Ratu
Date: 29 February 2012 14:12

Hello,
I have a question about .mtz files used in model building.
Here is how I did:
Diffraction  data - HKL 2000: .sca
CCp4i: scalepack2mtz: .mtz (1)
Phaser: In: template pdb & .mtz(1)
            Out: model .pdb(1) & .mtz(2)
Refmac5: model .pdb(2) & .mtz(3)
Here is the question:
1. Coot check and refinment: which mtz file shoudl I use?
2. With further refinemnt by refmac5, which mtz file should I use?
3. When I deposit data, which mtz file to use?
4. What is the difference between .mtz(1) and the .mtz files generated from "phaser" and "refmac"?
Thank you for advice
Ros

----------
From: Vellieux Frederic
Hi,

The refinement program generates an mtz file with map coefficients (including difference Fourier coefficients) so you should use that one for model rebuilding in coot;
for the next refinement rounds, at the beginning of each round you should provide the initial mtz file coming from data processing and post processing (1 in your terminology I suppose). This is because several refinement programs scale the Fobs (and sigmaFobs). Refmac is one of them I think;
for data deposition, you deposit the same mtz file (1), and if you also wish to deposit map coefficients you can also extract the map coefficients from the "final" refinement and map calculation round mtz (using, say, sftools) and deposit those.

Phaser does not carry out model refinement per se (adjusting atom positions and temperature factors). But the map coefficients present in the mtz file are generated in the same way (Sigmaa coefficients) with Phaser and with Refmac. There might be small difference in terms of the program code used internally but that shouldn't make much of a difference.

Since a model that has "seen" refmac is (normally) improved, the electron density maps generated using the refmac mtz should be improved wrt the maps coming directly from Phaser (molecular replacement). Normally, the model you refine will have had the differences in the sequence between molecular replacement search model and 'your' structure corrected (gradually?), the solvent model is introduced and improved, ligands, ions etc are being introduced... Thus the maps improve seen that your model should reflect more and more what is present in the crystal as you build and refine.

HTH,

Fred.

----------
From: Uma Ratu
Thank you very much!
It is clear for me now.
Cheers
Ros


----------
From: Eleanor Dodson
mtz(1) will contain h k l F SIGF I SIGI and optionally F+ SIGF+ and F- SIGF- This is your master file, providing the space group is correct

It may NOT have the correct space group however - MR searches may select one space group from several possible ones. If the final SG is different from the one specified after scalepack2mtz just run a jiffy to change it.

e.g. mtzutils hklin1 scal1.mtz hklout scal2.mtz
symm SGname
end

then use scal2.mtz as your master..

The output from phaser will have the same F but after anisotropic scaling. There will be other columns suitable for input to coot

the output from REFMAC will also give a scaled F plus extra columns .

Always start each new refinement with your master mtz as input..

coot will automatically select the best output from PHASER or REFMAC to calculate maps. The columns FWT and PHWT are used to generate a 2mFobs-DFcalc map The columns DELFWT and PHDELWT generate a mFobs-DFcalc map

Eleanor
--
Professor Eleanor Dodson


----------
From: Randy Read
Hi,

Just one correction to this.  Phaser doesn't substitute values with the same name in an output MTZ file.  To get the anisotropy corrected F and SIGF values, you have to run phaser in the separate anisotropy correction mode, and even then it appends the string "_ISO" to the column names and leaves the original columns as they were.  We make it hard to get the corrected values, because we think that downstream programs should be able to do a better job once they have more information (e.g. a well-refined structure), but we make it possible to get them because not all potential downstream programs are capable of doing any anisotropy correction at all.

Regards,

Randy
------
Randy J. Read


----------
From: Uma Ratu
Many thnaks for your input.
regards
Ros


----------
From: Maheshwaran amanthakadavu
Dear crystallographers,

According UMA RATU's mtz labellings,

      
Vellieux Frederic sir answered, output of the scalepack2mtz is used as master mtz for all refining (refmac5) input(if i am wrong, sorry ). am confused when Eleanor sir answered...

here my labellings !!!
 

scalepack2mtz ---- output_1.mtz

Phaser ----- output_2.mtz

Refmac5 --- output_3.mtz 
(output mtz from the refmac5 after first refining, input is phaser
output_2.mtz )


As
Eleanor sir told in second paragraph of his reply, i got  other best possible space group by running Phaser other than scalepack2mtz space group ( i mean output_2.mtz,C2221(Phaser) from output_1.mtz, C222 (scalepack2mtz) ) ..

Am right-now using output_2.mtz, Is it right to use output_3.mtz  ?


and

from Eleanor -- "scalepack2mtz just run a jiffy to change it."
how to do? am using ccp4 6.0.0 in  windows xp

Regards

VMW
--
----------------------------------------------------------------------------------------------
V . MAHESHWARAN


----------
From: Eleanor Dodson
Sorry my answers were confusing..

There are two problems. 1) When you finish data processing you should have the correct point group, but you may not have the correct spacegroup (SG). You will get an mtz file after processing, but may have to correct the SG later.

Eg an orthorhombic point group will have all angles 90 degrees and 2-fold symmetry around the a, b and c axis.

However the space group might be one of these 8. P 2 2 2, P 21 2 2 P 21 21 2, P 21 21 21, P2 21 2, P2 21 21, P 21 2 21 or P 2 2 21.

You can get an indicator of the SG from the systematic absences, but you may only be sure of it when you have solved the MR search or experimental phasing.

At that point you should go back and make sure the mtz from the data processing has the correct SG in the header.


2) After any sort of manipulation your output amplitudes in the mtz file will have been modified in some way.. This file will besuitable for reading into COOT for model building.

You should NOT use this file for later refinement though .
Eleanor


On Mar 14 2012, Maheshwaran amanthakadavu wrote:

Dear crystallographers,

According *UMA RATU*'s mtz labellings,

     *Vellieux Frederic *sir answered, output of the scalepack2mtz is

used as master mtz for all refining (refmac5) input(if i am wrong, sorry ).
am confused when Eleanor sir answered...

here my labellings !!!

scalepack2mtz ---- output_*1*.mtz

Phaser ----- output_*2*.mtz

Refmac5 --- output_*3*.mtz (output mtz from the refmac5 after first refining, input is phaser output_*2 *.mtz )


As *Eleanor* sir told in second paragraph of his reply, i got  other best

possible space group by running Phaser other than scalepack2mtz space group
( i mean output_*2*.mtz,C2221(Phaser) from output_*1*.mtz, C222
(scalepack2mtz) ) ..

Am right-now using output_*2*.mtz, Is it right to use output_*3*.mtz  ?


and

from* Eleanor -- "*scalepack2mtz just run a jiffy to change it."

----------
From: Rajesh kumar

Mahesh,

You should always use output_1.mtz for refinement. 

Dont use output_2. mtz for further refinement instead use some ccp4 utilities to change output_1.mtz (C222) to say output1_mod.mtz (C2221). This will become your master file for all further refinement. What ever remaining 2.mtz, 3.mtz, so on you get would only be used in coot and after the modification in coot the new model will be refined against your output1_mod.mtz (one with correct SG, now called new master file). 

Never use 2,3, 4, mtz files which you get after each round of refinement in next refinement. 
So "Am right-now using output_2.mtz, Is it right to use output_3.mtz  ?" - NO

Changing from 1 to 1_mod could be done in HKL2000, at scaling step or maybe reindex or pointless will do. I am not 100% sure though.  May be make sure about this as it says reindex because only you need to rescale it like in HKL2000. You could use SFtools and CAD also I guess. These three in all CCP4. There may be few more in CCP4 on windows, others could give good suggestion than me.

Hope this helps
Raj
  



----------
From: Rajesh kumar
This sentence in my previous posting may or may not be correct technically- May be make sure about this as it says reindex because only you need to rescale it like in HKL2000.  It its wrong I am  Sorry in advance.



1 comment:

  1. This is a very good question and always confuse beginners for crystallography. Thank everybody's effort to answer it. May I ask why we should always use the output_*1*.mtz, instead of output_*2*.mtz, as input of refinement? Many thanks in advance!

    ReplyDelete