xmipp_tomo_extract_subvolume (v3.0)


Extract subvolumes of the asymmetric parts of each subtomogram This program works closely together with the ml_tomo program, which is used to align (and classify) a series of subtomograms. Now, if each subtomogram contains some sort of (pseudo-) symmetry, tomo_extract_subvolume can be used to extract subvolumes of the asymmetric parts of each subtomogram. The metadata file that is output by this program orients each subvolume and its missing wedge in the correct orientation, so that it can be fed directly back into ml_tomo. There, the sub-subtomograms can be further aligned and/or classified


-i, --input <input_file>
Input file: metadata, stack, volume or image. Metadata file with input volumes (and rotations/shifts)
--mode <mode=overwrite>
Metadata writing mode.
where <mode> can be:
  • overwrite Replace the content of the file with the Metadata
  • append Write the Metadata as a new block, removing the old one
--label <image_label=image>
Label to be used to read/write images.
--oroot <rootname=>
Rootname for output stacks of subvolumes
-o <filename=>
Name of output metadata ("oroot".xmd by default)
--sym <sym=c1>
Symmetry group
--size <dim>
size output subvolumes
--mindist <distance=-1>
Minimum distance between subvolume centers, usefull to avoid repetition of subvolumes place at simmetry axis If set to -1 minsdist will be size/4
--center <x> <y> <z>
position of center of subvolume to be extracted

Examples and notes

Extract 12 vertices (subvolumes) in boxes of size 21x21x21 pixels from each subtomogram in the data set:
xmipp_extract_subvolume -i align/mltomo_1deg_it000001.doc -center 0 0 59 -size 21 -sym i3 -o vertices
Imagine we have N subtomograms of a virus particle and we have aligned them with respect to a reference structure (or we have aligned them in a reference-free manner). Then, suppose we are interested in the vertices of the virus: perhaps because we suspect that one of them is 'special': e.g. it is a portal. All we have to do is: * Pass the metadata of the aligned subtomograms of the ml_tomo prgram * Find the x,y,z coordinates of one of the vertices in the average (reference) map * Pass the (pseudo) symmetry description of the virus (icosahedral, see Symmetries) The program will then extract all symmetry-related vertices in each of the N aligned subtomograms. Each icosahedral has 12 vertices, so if the x,y,z coordinates coincided with a vertex (with an accuracy given by the --mindist option, see below) , then we will end up with 12*N sub-subtomograms. The sub-subtomograms are not rotated to avoid interpolation. Instead, the output metadata holds the correct orientations and translation for each of them to subsequently superimpose them in the ml_tomo program. Also the missing wedges of all sub-subtomograms will be treated correctly in this way. Then, in the ml_tomo program one could attempt to classify the 12*N vertices, in an attempt to 'fish out' the N supposed portal structures. The format of the input and output files is as described for the ml_tomo program. First, lets align our set of virus subtomograms (images.sel, with missing wedges defined in images.doc and wedges.doc, see ml_tomo for details) . We will use a known virus structure (myreference.vol) as reference and will use maximum cross-correlation (rather than maximum likelihood) to avoid problems with absolute greyscales and the standard deviation in the noise.
 ml_tomo -i images.sel -doc images.doc                      -ref myreference.vol --oroot align/mltomo_10deg -missing wedges.doc -iter 1 -ang 10
 ml_tomo -i images.sel -doc align/mltomo_10deg_it000001.doc -ref myreference.vol --oroot align/mltomo_5deg  -missing wedges.doc -iter 1 -ang 5 -ang_search 20
 ml_tomo -i images.sel -doc align/mltomo_5deg_it000001.doc  -ref myreference.vol --oroot align/mltomo_2deg  -missing wedges.doc -iter 1 -ang 2 -ang_search 8
 ml_tomo -i images.sel -doc align/mltomo_2deg_it000001.doc  -ref myreference.vol --oroot align/mltomo_1deg  -missing wedges.doc -iter 1 -ang 1 -ang_search 3
Now, we will identify the x,y,z coordinates of the vertex in the reference structure (which has symmetry i3): -center 0 0 59. And we will extract 12 vertices (subvolumes) in boxes of size 12x12x12 pixels from each subtomogram in the data set: xmipp_extract_subvolume -doc align/mltomo_1deg_it000001.doc -center 0 0 59 -size 21 -sym i3 -o vertices his has generated subvolumes in the same directory as the original subtomograms. For each subtomogram 12 subvolumes, which are named in the same way as the original subtomograms, but ending in _sub000001.vol etc. The orientations and translations are in a file called vertices.doc, the selection file is called vertices.sel. These files are already in the correct format for the ml_tomo program. Therefore, classifying the vertices into 4 classes using ML is now straightforward (see ml_tomo for details again) :
 xmipp_ml_tomo -i vertices.sel -doc vertices.doc -dont_align -keep_angles -missing wedges.doc -nref 4 -reg0 5 -iter 25
Note that the whole functionality of ml_tomo can be employed on the sub-subtomograms: alignment, classification, local angular searches, symmetry, etc. For example, in case of the vertices, on could apply -sym c5, and/or one could allow the vertices to re-orient slightly (with respect to the i3 symmetry) by using -ang 5 ang_search 15.

User's comments