On Thu, 25 Jun 2009 11:39:29 -0700 James Holton <jmhol...@lbl.gov> wrote:
> A long time ago, I wrote a little program for turning spot picks into > a PDB file that you could load up and rotate around in a graphics > program (such as O, at the time): > http://bl831.als.lbl.gov/~jamesh/pickup/spt2xyz.com > this one works on the *.spt files that comes out of MOSFLM. You can > concatenate *.spt files for each of your images to get the full data > set represented. > I also wrote one that works on ADSC images if you have the DPS > program installed: > http://bl831.als.lbl.gov/~jamesh/pickup/adsc2pdb.com > but it only works for the "common" coordinate system convention we > use at ALS and SSRL. Not sure about other beamlines. > > The algorithm for converting spot positions into reciprocal space > is not exactly new, as it is at the core of any and every > autoindexing program. All you do is figure out the take-off angle of > the diffracted ray from the crystal and then calculate the coordinate > where this line intersects the Ewald sphere. Specifically, the > "distortion" is: distortion = lambda*sqrt((Xdet)^2+(Ydet)^2+(dist)^2) > where: > lambda - the x-ray wavelength in A > Xdet - X coordinate of the spot relative to the beam center (in > mm) Ydet - Y coordinate of the spot relative to the beam center > (in mm) dist - the crystal-to-film distance (from crystal to > direct-beam spot, in mm) > > You then use this "distortion" to compute the x-y-z coordinate of the > reciprocal-lattice point in reciprocal space: > x' = Xdet/distortion > y' = Ydet/distortion > z' = dist/distortion - 1/lambda > > Yes, there are x-y-z coordinates in reciprocal space. They all have > units of inverse Angstroms. Of course, these x',y',z' coordinates > are at the phi value of the image you picked spots on. To get the > coordinates at phi=0, you need to "un-roate" them: > x = x'*cos(phi)-z'*sin(phi) > y = y' > z = x'*sin(phi)+z'*cos(phi) > > where some people can probably tell that in this convention the phi > axis is right-handed and parallel to the "Ydet" axis of the detector. > > The last step is to multiply these x,y,z coordinates by some > reasonable scale factor and put them into a PDB file. Maybe put the > intensity in the B-factor column, so that you can color it. Then you > need to find a display program that can handle a million-atom PDB. > Does anyone have one of those? > > > Ideally, what one would like to do is make an electron density map > with pixel-to-pixel correspondence to the image. All you do is apply > the above formulas to each pixel, do a Lorentz-Polarization > correction, and then just render the data set as a map. > > Sounds like there are a couple of commercial programs that do this in > 2 dimensions. Problem with doing it in 3D is there are no programs > that can display an electron density map this big. In fact, I can't > even get CCP4 to write out map or mtz files bigger than 2 GB. I get > "filesize limit exceeded" errors (even though the filesystem can > handle large files). Is this a limitation of the 32-bit binaries? > Can anyone help me confirm that re-compiling CCP4 as 64-bit will fix > this? This error can be reproduced by trying to make a 2 A data set > for the largest unit cell in the PDB: > The limit on file sizes using fopen and the like for 32 bit systems is 2^31 bytes, unless you use fopen64 or use _FILE_OFFSET_BITS == 64 passed to the compiler (assuming you have LFS support, etc, etc): http://www.gnu.org/software/libc/manual/html_node/Opening-Streams.html#index-fopen64-931 on 64 bit machines, the file size limit is 2^63 bytes (this also depends on the file system type, just to make things even more complicated). Here's another easy way to test it: dd if=/dev/zero of=bigfile bs=1024 count=3145728 Hope this helps! -Tim -- --------------------------------------------------------- Tim Fenn f...@stanford.edu Stanford University, School of Medicine James H. Clark Center 318 Campus Drive, Room E300 Stanford, CA 94305-5432 Phone: (650) 736-1714 FAX: (650) 736-1961 ---------------------------------------------------------
