On 06/16/2014 04:26 PM, George Devaniranjan wrote:
> Hi everyone,
>
> I apologize at the beginning itself as I am new to using PHENIX/X-ray crystallography so this question might not make the most sense but I will try and explain it well.
>
> I cut out a part of a map for a PDB fragment using phenix map cut out density resulting in a MTZ file.
>
> Now I changed the fragment (changed its phi and psi externally) and want to calculate the RSR [real-space R -factor] for the model based on the old density (to check if I have made things worse or not).
>
> how can I go about this? I tried using phenix.refine couldn't get it to work.
>
> I have attached the MTZ fragment, the original PDB fragment and the modified PDB fragment.
> native: nativeFragment.pdb
> model: modelledFragment.pdb
> MTZ: cut_out_density.MTZ
>
> (I can see mine is worse but I want to quantify that "badness")
>

I'm sure phenix has tools to do what you want, but if you still have difficulty
and want to go back to G. Kleywegt's suggestions of using mapman, step by step
instructions are below.

First you need to understand that an mtz file is not a map-
a map is a list of the value of density at points, like pixels in a digital picture
but in 3D (voxels) instead of 2D.
The mtz file contains structure factors which are the Fourier transform coefficients
of the map. With modern computers the map can be calculated rapidly from the structure
factors, and many applications work directly with mtz files and never save a map file.
Still the distinction is important- for examples you cannot calculate a CC between
mtz files in a defined area, because of the all-on-all relation between
reflections and realspace. You can calculate the CC between two maps in a defined
area, which is what the suggested mapman command does.

As I understand it, mapman calculates the RSCC and RSR between two maps- the
pdb file is only used to define the area in which the values are calculated.
Since it uses the pdb to define the area, there is no need to cut out density
around the fragment- I recommend using the original mtz to calculate the "obs" map,
that way you can use the same one for all fragments.

So you need to calculate
a map from the original mtz covering all the fragments
a map (Fc-Phic) from your fragment
(They presumably must have the same cell parameters and grid spacing)
and give them together with the fragment.pdb to mapman.

Do this once for nativefragment.pdb
and again for modelledfragment.pdb
and compare the results
Remember phases in the original mtz wil be biased in favor of the original model,
so if the modified ragment is equally good it will still get a worse score.

step by step:

1. Use fft to calculate a map from the original mtz:
 From the ccp4i gui, on the left side, select program list, then under that fft.
fill in according to the screen shot at  http://sb20.lbl.gov/ccp4/RunFFT-1.png
(this is working with your cut-out-density.mtz but I recommend using the
original mtz). see notes below.

2. Use sfall to calculate SF's from nativefragment.pdb making nativefragment.mtz
	(http://sb20.lbl.gov/ccp4/RunSFALL.png)
	(with a script sfall can make the map directly, don't see that in the gui)

3. use fft to calculate a map from nativefragment.mtz
	(http://sb20.lbl.gov/ccp4/RunFFT-2.png)

4. run mapman in the directory where these files were created:

MAPMAN > read m1 cut_out_density.map ccp4
MAPMAN > read m2 nativefragment.map ccp4
MAPMAN > RS_fit
  Which obs_map ? (m2) m1
Which calc_map ? (NATIVEFRAGMENT.MAP) m2
  Model PDB file ? (m1.pdb) nativeFragment.pdb
  RS-fit list output file ? (rs_fit.list) fragment27CC.txt

 . .
  Non-water residues with RSCC : (         28)
  ... Average RSCC for these   : (   0.775)

  Non-water residues with RSR  : (         28)
  ... Average RSR for these    : (   0.247)

6. now repeat 2 - 4 using modified fragment instead of native fragment


Notes-
FFT-1:
  Uncheck the box that says "append to an existing . . ."
FFT will probably guess which columns to use, at least if coot was able to auto-open.
if this is blank and you don't know, send a list of the column labels.
set the resolution say 100 to 2 A.
click the bar that opens up "infrequently used options". You need to
set the grid spacing so it will be the same in both maps, and fine
enough that it doesn't distort either map. Should have no large prime
factors- power of 2 like 64, 128 etc is fine.
(Or let sfall choose grid based on resolution, note what it is and use the same
in step 3 when you make the other map)
If your original mtz is not P1 and all of the fragments are not in the asymmetric unit,
you may need to use the option to extend map to cover (whole) pdb
You only need to do this once in that case, for all the fragments.

sfall:  make sure the cell param are the same as in step 1, which will be determined by orig mtz.

fft-2: use the same grid as in step 1. cell param taken from the mtz should be right