Hi Christopher,
Ok, I see, you convert your MRC map to BRIX for working in O but then
bring the PDB model that was adjusted in O back into Chimera and realign
it with the original MRC. It sounds like the painful part of this
process is that the BRIX map does not align with the original MRC map
when viewed in Chimera. That is a problem with the SPIDER conversion
from MRC to BRIX losing or interpreting differently the origin
information. If the BRIX map actually had the correct origin so it
aligned perfectly with the MRC map in Chimera then it would save you a
bit of trouble.
One solution is of course to have Chimera write out the BRIX file so
it gets the positioning correct. I'm not sure when I would find time
to write that code. Perhaps another solution is to figure out how the
SPIDER conversion messes up the origin. That might allow choosing an
MRC origin in Chimera such that the SPIDER BRIX map will align with the
MRC map. My guess is that Chimera is using the MRC 2000 xorigin,
yorigin, zorigin header fields while SPIDER is using the ncstart,
nrstart, nsstart header values which Chimera sets to zero when it writes
an MRC 2000 format file. If this theory is correct, the BRIX map might
align with the MRC map if you change the MRC origin to (0,0,0) using the
Chimera volume dialog Features / Origin and Scale panel. You could save
this MRC map and work with it to avoid the BRIX/MRC misalignment problem.
Tom
Christopher Akey wrote:
> Tom-
>
> As to our problem with getting files back into O without degradation of
> information. Here is what we found that works.
>
> You have an original image in mrc format(1.mrc)=> convert to brix (we
> use SPIDER)
>
> Then in O one can dock pdbs & do modeling etc. So now one has a brix
> file and pdb modeled in approximately the right area. However, you want
> to see if you can improve this difficult modeling.
>
> So you take the brix file and the pdb into Chimera.
>
> You write out the brix file as a new mrc file, ie 2.mrc, without any
> transformations.
>
> You then read in the original (1.mrc) and move 2.mrc relative to 1.mrc
> using local movements with mouse and then the the map vs map fitting.
>
> You then transform the fitted pdb file into the original 1.mrc using
> "transform as" following the changes to the 2.mrc file.
>
> You can now do a zone or zone followed by a step of volume erase around
> the pdb file using the 1.mrc file, they are in the same relative
> orientation now.
>
> Finally, you write out a zoned/ve map, 1_zone_pdb.mrc, using the 1.mrc
> map, so no loss of resolution here.
>
> Then you can convert 1_zone_pdb.mrc to a brix file and look at in O with
> the pdb and it will now be docked in the correct way, ie one won't have
> lost the correct origin/alignment of the pdb to the zoned map.
>
> ------
>
> At this point in O, one can adjust the modeled pdb in the zoned density,
> then bring the brix file and new modeled pdb files back into Chimera to
> make a figure with a better fit of the modeled pdb file.
>
> This is obviously only needed when you have two structures interacting
> and need to segment them vary carefully, to try and improve the local
> fit of a loop or whatever.
>
> Did this make sense?
>
> It is a way to work in Chimera then transfer back
> into O to visualize and work there and even to cycle between them.
>
>
> cheers C Akey
