Just a detail: you can switch off the map calculation in phenix.refine
to suppress the huge & unnecessary maps with
refinement.output.write_maps=False
Phil
On 2 Jul 2008, at 09:31, Kevin Cowtan wrote:
There are some issues which may be affecting some people using Coot
with PHENIX. If you try and use the two together in the wrong way,
then you may get unexpected results. (Problems hashed out with help
from Dirk).
phenix.refine produces two different sets of output files. I quote
from the documentation:
At the end of refinement the program generates:
...
two maps: likelihood weighted mFo-DFc and 2mFo-DFc. These are
in ASCII X-PLOR format. A reflection file with map coefficients
is also generated for use in Coot or XtalView (e.g.
lysozyme_refine_001_map_coeffs.mtz);
There are some CNS maps, which may be used in conventional graphics
programs, and there is an MTZ file, which may be used in Coot or
XtalView.
The difference is important. Coot and XtalView have a fundamentally
different concept of what a map is to other graphics programs. As a
result they require different information. Phenix.refine recognises
this and provides appropriate information for each type of program.
In the last year we incorporated the ability to read CNS files into
coot. As a result, some users have started reading the CNS map files
into coot instead of the MTZ file. Unfortunately, this does not
always work the way you would expect, because the map files has not
been generated with Coot in mind. This may lead to misleading sigma
levels, or, if the map is modified or masked in any way before
giving it to Coot, much more serious artifacts.
The solution is simple: When using Coot, use the MTZ file which the
phenix developers have provided for that purpose, *not* the CNS map
files. In addition to giving the right answers, there are some other
benefits:
- The MTZ is much smaller, in case you want to archive it.
- Using auto-open mtz, you get both maps at once, and the difference
map is automatically displayed as such, saving mouse clicks.
- The spacegroup and cell information are in the file, so it works
even if you don't open the PDB file first.
..
You can stop reading here, the rest is detail for those who are
interested:
Coot and XtalView use a different representation of electron
density, a continuous crystal representation. Most other graphics
programs use a bounded box representation.
The continuous crystal representation means that the electron
density is everywhere, it is omnipresent. "If I scroll up to the
heavens, density is there; if I place my ligand in the depths,
density is there. If I drag over to the far side of the solvent,
even there the density will guide my fitting." (*)
To generate a continuous crystal representation, you need a complete
and consistent description of the electron density over the
asymmetric unit (any asymmetric unit - it doesn't matter where in
space). The bounded box representation need the electron density
over the volume of the molecule.
It is not always possible to get the asymmetric unit representation
from the bounded box representation, because in creating the bounded
box some information may have been lost. However, you can always get
it from an MTZ file, or from a map file on the condition that the
map file is appropriately bounded.
Kevin
(*) Psalm 139 v8-10
