Using the Batch Reconstruction Interface with Dual-axis Tutorial Data Sets

(IMOD 4.11)

University of Colorado, Boulder

This document will take you through the steps required to reconstruct the IMOD dual-axis tutorial data sets in Etomo's interface for batch reconstruction with Batchruntomo. It explains some of the most important points as they arise. For more details, consult Batch Tomography Guide, which you should read through before trying to process your own data sets in this interface. In this document, labels in the interface are shown in Bold, and entries in fields are shown in Italics. 

Get started:

• Download both the binned and the unbinned tutorial data sets from the IMOD web site; this allows more features to be illustrated than doing just one data set.
• Move both files, "tutorialData.tar.gz" and "tutorialData-1K.tar.gz", to the directory where you want to do batch processing.
• In a terminal, cd to that directory.
• Unpack the unbinned set, rename its directory, and unpack the binned set:

      imoduntar tutorialData-1K.tar.gz
    mv tutorialData BB-unbinned   (Use ren instead of mv on Windows without Cygwin)
    imoduntar tutorialData.tar.gz

Go to the batch interface and set up the batch parameters:

• Start Etomo with
     etomo
• Press Batch Tomograms on the Front Page.  The interface opens with the Batch Setup tab.
• Select plasticSection.adoc as the System Template. This template turns on Sobel filtering when tracking beads to give improved centering of the points.
• The initial radio buttons allow stacks to be moved from elsewhere into automatically created data set directories.  Your stacks are already in their own data set directories, so there is no need to change the default selection.
• At the bottom, two fields allow one to change the root name that will be used as a prefix for all the batch processing files, and to select a different Location in which various batch project files will be kept.  Again, there is no need to change the defaults here.

Add the two stacks, unbinned then binned, and set their properties:

• Switch to the Stacks tab.
• Press Add Stacks.
• Browse to the "BB-unbinned" folder and select "BBa.st". Notice that the interface detects that the set is dual-axis.  You could also select both "BBa.st" and "BBb.st", and the interface would include only one data set.  This feature is handy if you need to select several dual-axis sets in one directory.
• Check Beads on Two Surfaces
• Press Add Stacks.
• Go to the "tutorialData" folder and select "BBa.st". The set inherits the Two Surfaces setting from the previous entry.
• Notice the right arrow button in the Stack column header. Press this to show the full paths of the files and see that you have files in two different directories.

Set parameters for the data sets:

• Switch to the Dataset Values tab. This "Basic" screen presents a carefully selected subset of the many parameters shown on the 9 different screens in the Reconstruction interface.  Parameters are provided to Batchruntomo in text files containing "directives".  Other options can set in an "Advanced" screen opened by pressing the Advanced button, or by supplying them in a template file or a starting batch directive file.
• Select Remove X-Rays.
• Enter 10 for Bead size (nm) and 25 for Target number of beads. (There are actually 28 beads of high quality in these data sets; others are paired or elongated. Enter 28 if you wish.)
• Select Enable distortion (stretching) in alignment and Use local alignments. These selections do not guarantee that these alignment features will be used. Stretching will not be used if there are not enough beads on both surfaces; local alignments also require a minimum number of beads.
• Select Do positioning for and leave the radio button set for Plastic section. There is no need to enter a tomogram thickness since the automatically selected thickness will work.
• In the Reconstruction section, the selection of the Calculated thickness is appropriate; the thickness will come from the positioning step, but a thickness would be available from the positions of the gold beads on two sides even if the positioning step were not included. However, an entry for With fallback is required with this choice in case both of those sources of information come out with too low a thickness. Enter 250 here. With a pixel size of 1 nm for a 300 nm section, the unbinned data set is expected to have a thickness of about 180-200 pixels.
• Also enter 16 for Plus (optional); this will allow some extra thickness that can be trimmed away in the postprocessing step.
• Select Postprocess with trimvol. This will cause the volume to reoriented with rotation around X. Turn on Find plastic section limits and add and enter 10 to find the limits of the section and trim to 10 pixels more than that. Turn on Fraction of Z slices to analyze to convert to bytes; the default fraction of 0.33 is fine here so the field can be left blank.
• To give the binned data set some different values, press Open on the second line of the Datasets table. A separate Dataset Values screen will open for this data set.
• Turn off Remove X-rays, which is not appropriate for the binned data. In the Reconstruction section, set With fallback to 130 and Plus (optional) to 10. Open Postprocessing and set the added pixels to 6.
• For more variety, you can turn on Erase gold, select Find beads in 3D, and set the Tomogram thickness (pixels) to 130.
• Press OK to close this secondary dialog. Notice the "Set" entry in the table to indicate that this set has its own values.

Switch to the Run tab to run the data sets.

• The initial default is to use one CPU and no GPU. Once you select Use multiple CPUs, Parallel GPUs, or both, the Resources table will appear at the top. There, depending on your computer setup, you can select how many cores to use, which machines to use them on if you have access to multiple machines, and what GPUs to access. A single GPU is quite sufficient for these simple data sets.
• Make whatever selection of CPU and GPU resources that is appropriate for your computer.
• Press Run. Directive files will be written for each data set and the processing will begin. Watch the status output in the table at the bottom. Pausing, killing, and resuming are possible but more complicated than for ordinary processing; see Running the Data Sets in the Batch Guide for details.
• Essential items like mean residual values and warnings appear in the project log during the processing. At any time after a set begins being processed, you can see the full output from the batch program by pressing the button in the Open Log column. Once a set has finished or reached a stopping point, it can be opened in the Etomo reconstruction interface with the button in the Open Set column. Once there is a final reconstruction, it can be opened with the 3dmod button in the Open Rec column.
• You can see that if you turned on Stop after, the default choice would be to stop after the gold detection, which might be useful if you were erasing gold. Both data sets would run to that point for both axes (regardless of whether those steps were actually run for a particular step). When you restart, both sets would restart after that point, unless an earlier point were selected.