Tutorial - Tomographic Reconstructions in Priism (1/2)

This tutorial will guide you through the reconstruction process. There is a cryo dataset (473 megabytes) available as an example. Start by downloading the dataset and taking it through the entire process. This page only covers the first part of the tutorial; for the second part, visit http://msg.ucsf.edu/em/EMNEW2/Tomog_tutor_v2_2.html.

1. Assemble the Raw Data

UCSF_TOMO outputs an ordered stack of data starting from the most negative tilt through to the most positive tilt (Movie 1). Contained within the header of this file are the tilt angles and the tilt axis and shifts used to collect the data. Note, the values of the tilt axis and shifts, while quite good, may need to be further refined as described below. If your data is coming from some other source, you can use the programs CopyRegionand insert_tilts to reorder and put the proper tilt angles in for the rest of the process. You can use Merge_reorder in place of CopyRegion. As a side note, the example dataset was preprocessed with CopyRegion to throw away a 24 pixel wide border on each side that contained camera artifacts; such preprocessing should not be necessary with any datasets recently collected on the microscopes in the Keck Advanced Microscopy Laboratory.

Movie 1. This movie shows a Drosophila centrosome imaged at 300kV, +/- 60 degree tilt range with projections every 2 degrees, and a 20eV slit. Total electron does is 200 electrons per square angstrom. Note at one point the slit becomes visible. The QuickTime movie can be downloaded from http://msg.ucsf.edu/em/EMNEW2/movies/Cent_001mrc_v2.mov. It is 7.8 megabytes long.

2. Append Lower Resolutions

We usually being working with a reduced size version of the dataset, say 512 x 512, rather than the full size of the collected data (2048 x 2048 when the data is collected with a binning factor of two on a 4096 x 4096 camera) to reduce processing time and to improve the signal to noise ratio for the cross-correlation alignment. The most convenient way to work with different versions of the dataset at different resolutions is to use Priism's AppendRes (from the Priism menus, first select "DataViews", then "Resolution", and finally "AppendRes). To use AppendRes, follow this recipe:

  1. In the AppendRes dialog, press the "Dataset" button to open a file browser to select the input tilt series; alternatively, you can enter the path to the file in the text field immediately adjacent to the "Dataset" button.
  2. Enter the number of resolutions you want in the "# of Resolutions" field. For the example dataset, you'll want to enter three in the "# of Resolutions" field (one for the original data, one for a 1000 x 1000 version, and a final one for a 500 x 500 version).
  3. Enter one in the "Z Scaling" field because you do not want the lower resolutions to combine projections at different tilt angles.
  4. Press the "DoIt" button at the bottom of AppendRes's dialog to modify the file.

Image 1. Shows a screenshot of Priism and AppendRes graphical user interface with the AppendRes parameters appropriate for this example. screenshot

3. Use EMTAR for the Initial Alignment.

The EMTAR dialog (in Priism's menus, select "Processing", then "EM Processing", and finally select "Alignment + Reconstruction") is the graphical interface you will use for aligning and reconstructing the tilt series. That dialog provides a way to run the different steps in the process either one at a time or several at once in a fixed order. This tutorial will use the steps one at a time: that is convenient for learning what the steps do and for checking on the results before proceeding with subsequent steps. Once you are familiar with the stages and have some confidence that the selected parameters are appropriate for your data, you turn on multiple stages and run them at once.

For move information about EMTAR beyond what this tutorial provides look at http://msg.ucsf.edu/IVE/IVE4_HTML/EMCAT/EMTAR.html.

Much of the process is computationally expensive. For the reconstructions, you can speed them up by using a machine with multiple processors or a cluster of several machines: see the notes on parallel processing. For the Macromolecular Structure Group at UCSF, the clusters, ec3 and ec6, are available for performing reconstructions.

For the sample dataset, these steps will extract the alignment parameters computed by UCSF_TOMO:

  1. In the EMTAR dialog, use the "TiltSeries" button to bring up a file browser and select the tilt series file; alternatively, enter the path to the file in the adjacent text field.
  2. Enter the resolution you want to use for calculations in the "Resolution" field about halfway down EMTAR's main dialog. For the sample data set, enter two in that field to select the second lower resolution (500 x 500 in this case). The alignment parameter results will be in the full resolution coordinate system to make it easier to reuse those results when reprocessing at a higher resolution. Other results, like the aligned and vvv mass-normalized images or the final reconstruction volume, will be at the lower resolution. Size-related parameters in the EMTAR dialogs are in full resolution units
  3. Press the small button immediately to the left of the larger "Rough alignment" button. That turns on the rough alignment stage of processing and opens the parameter dialog for that processing stage. In this case, the default parameters (which are to extract the alignment parameters from the extended header) do not need to be modified and you can close the dialog that appears. If you want to see that dialog again without turning on or off the rough alignment processing stage, press the "Rough alignment" button in EMTAR's main dialog. Note that some controls are toggle buttons which are either on or off; if a toggle button is on it will appear darker with a light shadow on one side (to make it look like the button has been pushed inwards).
  4. Press the "Do It" at the buttom of EMTAR's main dialog. That runs all the processing stages you've selected (in this case that is only the rough alignment). The rough alignment for this example will generate Cent_ufix_001.bprmMn (along with a log file containing notes about the calculation and the script file used to perform the calculation) in the same directory as your input tilt series. Cent_ufix_001.bprmMn contains the shifts used by UCSF_TOMO to track the sample during data collection. If your data did not come from UCSF_TOMO, the rough alignment will compute a cross-correlation based alignment: in order for those calculations to work well you should enter your best estimate of the orientation of the tilt axis in the rough alignment's "Axis orientation" field. To see that field, open the parameter dialog for the rough alignment and then press the toggle button next to "EMROUGH: special parameters"; the "Axis orientation" field will be at the top of the dialog that appears after you press the toggle button.
  5. Turn off the rough alignment step in the EMTAR dialog by pressing the small button immediately to the right of the "Rough alignment" button. Leave the EMTAR dialog open; you'll be using it for the later steps in this tutorial.

Image 2. Shows a screenshot of the EMTAR dialog with the rough alignment stage enabled and the rough alignment parameter dialog with the default parameters. screenshot

4. Use EMTAR to Compute Mass-normalization Parameters

To compute the mass-normalization parameters for the sample dataset:

  1. Return to the EMTAR dialog you had opened for the rough alignment and turn on the mass normalization processing step.
  2. In the mass-normalization parameter dialog, change the menu next to "Quantity to fit" to "average intensity" rather than "background intensity". Look at documentation for the mass-normalization step if you want more information about when to use the average intensity rather than an estimate of the background intensity level as the basis for the mass-normalization calculations.
  3. In the special parameters dialog for the mass-normalization, change the menu that reads "save only" to be "show and save". That will cause the mass-normalization calculations to automatically display diagnostic graphs. Once the mass-normalization calculations complete, you can also use the "MASSNORM fit" option from the "Display" menu in EMTAR's main dialog to view those graphs.
  4. Press the "Do It" button at the bottom of EMTAR's main dialog. The alignment parameter file, Cent_ufix_001.bprmMn, will be updated with the calculated mass-normalization parameters and you should see a graph of the mass-normalization fit (Image 4 shows a screenshot of that graph).
  5. In the EMTAR dialog, turn off the mass-normalization step.

Image 3. Shows a screenshot of the EMTAR dialog with the mass-normalization stage enabled and the mass-normalization parameter dialogs. screenshot

Image 4. Shows a screenshot of the diagnostic plot for the mass-normalization fit. screenshot

Part 2 of the tutorial