saving maps generated with molmap - origin?
Hello, I'm wondering why when I generate a density map from a pdb structure (using the molmap command) and then save it, once I reopen it it seems to have shifted with respect to the structure. I'm noticing that before saving it, the generated map has a certain origin (shown in the Coordinates pane in the Volume Viewer), and then when I reopen it the origin appears as 0 0 0. If I copy the origin shown initially and then paste it into the origin of the opened file, this shifts the map to the correct position, but why isn't the right origin saved with the file? I'm using alpha version 1 build 2509 2008/05/01, platform win32 Many thanks, Greg
Hi Greg, The trouble is you are writing out the map in BRIX format which I believe only allows integer grid index values for the origin of the map. The molmap command creates a map whose origin is not at an integral grid index. I could be wrong about the BRIX format, it may allow floating point origin values, but the documentation for BRIX does not say and gives example code for writing the file which uses integer format. I guess the next step is to try writing a floating point origin in to the text BRIX header and open it in O and see what happens (error, misaligned, or correctly aligned). To avoid this problem you can write your map out in MRC format. Tom Grigore Pintilie wrote:
Hi Tom, Thanks for the quick response! Actually, and sorry for not specifying previously, I am writing the file to MRC format... From my understanding of this fromat, it also requires specification of the origin as an integral grid index (in the field nstart). I've now gotten around this problem by writing a little C++ program that reads the mrc file, changes the nstart field, and then writes the mrc file again. I put the correct origin, which i get from chimera right after generating the map, as a floating point number, in the variable o, and then in the C++ program change nstart as follows: nstart[0] = (int) round ( o[0] / step[0] ); nstart[1] = (int) round ( o[1] / step[1] ); nstart[2] = (int) round ( o[2] / step[2] ); (step is the variable that contains the dimensions of the grid cells) this works perfectly, when i load the mrc file again, it has the correct position and origin. It's rather strange that chimera doesn't seem to save the origin properly. It would be nice to not have to call my C++ program, and also it might help others using chimera... although I can understand the dilemma with the origin specified as an integer... Greg On Tue, Aug 19, 2008 at 1:23 PM, Tom Goddard <goddard@cgl.ucsf.edu> wrote:
Hi Greg, I tested writing the molmap output to MRC format with the current Chimera production release 1.2540 (newer than the 1.2509 version you are using) and it gave the correct non-zero origin. Could you try with 1.2540? Chimera writes MRC 2000 format which uses floating point xorigin, yorigin, zorigin fields instead of ncstart, nrstart, nsstart integer origin header values. I'm not clear how your C++ fix that modifies the header is working, but I don't think it worthwhile for me to debug an old version of Chimera. If you can produce the problem in the current Chimera production release then I'll look into it. Tom Grigore Pintilie wrote:
I just installed the latest version and right away reproduced the behaviour I was describing. Here is what I was doing: when saving the map, I chose the MRC format, but then gave my file the extension .map When loading this file, the origin information is not loaded properly. But... I then tried again generating the map and saving it in the MRC format, but not entering an extension, hence defaulting to .mrc. Then when I load this file the origin is loaded correctly! Great! So it looks like the problem lies when loading a MRC file that has any extension .map. The map is loaded properly, but the correct origin is lost. (To test this, I saved a map as MRC using the extension .map, then changed the extension to .mrc, and surprise, the origin is loaded correctly!) Is there an description of the MRC 2000 format anywhere? I still seem to be able to load a map I saved as .mrc with my program, which uses the old format: http://www.ccp4.ac.uk/html/maplib.html#description Greg On Tue, Aug 19, 2008 at 1:52 PM, Tom Goddard <goddard@cgl.ucsf.edu> wrote:
Hi Greg, If you add the file suffix ".map" then Chimera will interpret the file as a CCP4 map which is slightly different than MRC format. The most important difference is that MRC has the floating point origin specification while CCP4 does not. You can open a file with ".map" suffix in Chimera as an MRC file by choosing "All (ask type)" for the File Type menu in the Chimera File / Open dialog. Or if you open the file from the Chimera command line use the prefix "mrc:", for example Chimera command "open mrc:myfile.map". Simplest just to use the ".mrc" suffix when you save the file. Look at section 5 of the following MRC documentation for the MRC 2000 format spec. http://www2.mrc-lmb.cam.ac.uk/image2000.html Tom
participants (2)
-
Grigore Pintilie -
Tom Goddard