We have an issue with disparity of the interpretation of CCP4 and MRC density maps in Chimera. The CCP4 and MRC file formats are very close in specification, and in particular the definition of the origin is the same. However, when I open both formats of a density map in Chimera, the origins given in the Volume Viewer do not agree and the maps are located in different places. The first issue is how the origin is specified. In the 3DEM community, we proposed that the origin is defined as the location of the physical coordinate origin within the framework of the map, expressed in voxel dimensions. In Chimera (build 2422), the origin for the MRC format is consistent with this definition while for the CCP4 format it is not. Example: I have a map with dimensions 100x100x100, an origin at 50x50x50 and sampling of 2 angstroms/voxel. In both the CCP4 and MRC versions the relevant fields are: nxstart = -50, nystart = -50, nzstart = -50. The origin as defined for 3DEM is thus the negative of these fields. In Chimera, the origin is given as {50,50,50} for the MRC format file, while for the CCP4 format file it is {-100,-100,-100}. Suggestion: The origin given in the Volume Viewer for the example should be either {50,50,50} specified as voxel units, or {100,100,100} specified in angstroms, for both file formats. Identical maps in the two formats should be superimposable when the origin specifications are the same. Bernard Heymann, Research Fellow Rm 1515, 50 South Dr., MSC 8025, NIAMS, NIH Bethesda MD 20892-8025 Tel. 301-451-8241, Fax. 301-480-7629
Hi Bernard, The Chimera origin reported in the Volume Viewer dialog is not the array index of physical origin. It is the physical coordinate of array index 0,0,0. So it is the inverse of how you are interpreting it. We've been using this definition since about 2000, before your 2006 paper on conventions for EM maps. With this understanding your example of a 100^3 map with physical origin at array index (50,50,50) and 2 angstroms/voxel the index (0,0,0) physical coordinate is (-100,-100,-100) which you say Chimera reports for a CCP4 map. Chimera uses the xorigin, yorigin, zorigin header values (words 50-52) in MRC2000 format instead of the nxstart, nystart, nzstart header values (words 5-7). You probably have (50,50,50) in the xyz origin fields while Chimera expects (-100,-100,-100) in those fields. The MRC documentation http://www2.mrc-lmb.cam.ac.uk/image2000.html unfortunately does not describe the meaning of those origin fields, though I suspect your interpretation is correct as it agrees with nxstart,nystart,nzstart definition and would just provide floating point equivalents of those integer values. Perhaps there is better MRC documentation defining xorigin,yorigin, zorigin. Chimera writes MRC2000 files and fills the origin fields with the physical coordinate of index (0,0,0). These different interpretations of the meaning of the origin fields are a big problem. How are the xorigin,yorigin,zorigin fields in MRC2000 actually being used? I know Chimera and EMAN both interpret it as a physical coordinate of array index (0,0,0). From your comments I assume BSOFT interpets it as the array index of the physical coordinate origin. What other software is using these fields and how do they interpret it? It would be a big problem to change the origin definition in Chimera because users have written many MRC2000 files using Chimera that would no longer be positioned correctly if the meaning of the origin fields is changed. Chimera does not write information in the MRC labels header fields indicating which version of Chimera wrote the file. That was a mistake. So there is no way of telling which interpretation should be used for a Chimera-written MRC file. EMAN currently writes a label line saying EMAN and the date and time the file was written. Tom
Hi Tom Thanks for the clarification. It turns out I had no idea how the new origin fields are interpreted. It is also not clear from the limited documentation how it should be interpreted and there are no example images to see how it is done. So I decided to just follow your explanation and fix it in Bsoft such that it displays correctly in Chimera. Thanks again Bernard On Oct 23, 2007, at 2:24 PM, Tom Goddard wrote:
Hi Bernard,
The Chimera origin reported in the Volume Viewer dialog is not the array index of physical origin. It is the physical coordinate of array index 0,0,0. So it is the inverse of how you are interpreting it. We've been using this definition since about 2000, before your 2006 paper on conventions for EM maps. With this understanding your example of a 100^3 map with physical origin at array index (50,50,50) and 2 angstroms/voxel the index (0,0,0) physical coordinate is (-100,-100,-100) which you say Chimera reports for a CCP4 map.
Chimera uses the xorigin, yorigin, zorigin header values (words 50-52) in MRC2000 format instead of the nxstart, nystart, nzstart header values (words 5-7). You probably have (50,50,50) in the xyz origin fields while Chimera expects (-100,-100,-100) in those fields. The MRC documentation
http://www2.mrc-lmb.cam.ac.uk/image2000.html
unfortunately does not describe the meaning of those origin fields, though I suspect your interpretation is correct as it agrees with nxstart,nystart,nzstart definition and would just provide floating point equivalents of those integer values. Perhaps there is better MRC documentation defining xorigin,yorigin, zorigin. Chimera writes MRC2000 files and fills the origin fields with the physical coordinate of index (0,0,0).
These different interpretations of the meaning of the origin fields are a big problem. How are the xorigin,yorigin,zorigin fields in MRC2000 actually being used? I know Chimera and EMAN both interpret it as a physical coordinate of array index (0,0,0). From your comments I assume BSOFT interpets it as the array index of the physical coordinate origin. What other software is using these fields and how do they interpret it?
It would be a big problem to change the origin definition in Chimera because users have written many MRC2000 files using Chimera that would no longer be positioned correctly if the meaning of the origin fields is changed. Chimera does not write information in the MRC labels header fields indicating which version of Chimera wrote the file. That was a mistake. So there is no way of telling which interpretation should be used for a Chimera-written MRC file. EMAN currently writes a label line saying EMAN and the date and time the file was written.
Tom
Bernard Heymann, Research Fellow Rm 1515, 50 South Dr., MSC 8025, NIAMS, NIH Bethesda MD 20892-8025 Tel. 301-451-8241, Fax. 301-480-7629
participants (2)
-
Bernard Heymann -
Tom Goddard