Practical Suggestions for Tomographic Reconstruction - Revision history

Stefano Scaramuzza at 13:59, 30 April 2019

2019-04-30T13:59:37Z

Stefano Scaramuzza at 09:33, 29 April 2019

2019-04-29T09:33:00Z

Stefano Scaramuzza at 13:19, 10 April 2019

2019-04-10T13:19:17Z

Stefano Scaramuzza at 15:55, 8 April 2019

2019-04-08T15:55:21Z

Stefano Scaramuzza: Created page with "This article contains some practical suggestions and examples for the pre-processing of tomographic tilt series and the reconstruction of tomograms. Be aware that there exist..."

2019-04-05T16:51:41Z

Created page with "This article contains some practical suggestions and examples for the pre-processing of tomographic tilt series and the reconstruction of tomograms. Be aware that there exist..."

New page

This article contains some practical suggestions and examples for the pre-processing of tomographic tilt series and the reconstruction of tomograms. Be aware that there exist many other working strategies and softwares in addition to the examples shown here.

=== Naming convention ===
When dealing with large and growing datasets, we highly recommend to strictly follow a consistent file naming convention. A possible convention is to use a number for the batch (a batch stands for one data acquisition session) and a number for the tilt series within this batch. E.g., the tilt series 21 of batch 1 would be named ''b001ts021.st''. The same goes for all other related file types such as ''b001ts021.rec'', ''b001ts021.fid'' and so on.

=== Organizing data ===
We recommend to organize all your data in a comprehensive and systematic way from the very beginning. One possible way of organizing is shown below.
[[File:data_organisation_dynamo.jpg|thumb|center|200px| Example of organizing your data.]]

=== Drift correction ===
Any software for drift correction should give good results. A wrapper for [http://msg.ucsf.edu/em/software/motioncor2.html motioncor2] can be found in the [https://github.com/C-CINA/TomographyTools TomographyTools].

=== Exposure filtering ===
If needed, the exposure filtering can be done independently from the drift correction using the function found in [https://github.com/C-CINA/TomographyTools TomographyTools].

=== Example of a pre-processing script ===
The pre-processing script from the [https://github.com/C-CINA/TomographyTools TomographyTools] takes the original 4k movies (''TS31_000_0.0.mrc'', …) and does the drift correction, the exposure filtering and the merging of the drift corrected movies into tilt series. The outputs are an exposure filtered stack (e.g., ''b001ts031.mrc'') and an unfiltered stack (e.g., ''b001ts31_UW.mrc'') that can be used later for CTF estimation.

=== Tilt series alignment ===
See [[Considerations_for_tilt_series_alignment_in_IMOD | considerations for tilt series alignment]] with [https://bio3d.colorado.edu/imod/ IMOD].

=== Tilt angle convention ===
We recommend to check that the tilt angle convention is correct in the tilt angle files. How and why this is done is explained [[Tilt angle convention | here]].

=== CTF estimation ===
The CTF is ideally estimated on the drift corrected stack (but not exposure filtered and not aligned). A matlab wrapper for [http://grigoriefflab.janelia.org/ctffind4 ctffind4] can be found in [https://github.com/C-CINA/TomographyTools TomographyTools]. If the CTF is estimated outside of the IMOD pipeline, the “Run Ctf Plotter” button from IMOD has to be skipped and the results from ctffind4 are used instead. The output from ctffind4 needs to be converted first into a file format that can be read by IMOD with the function found in [https://github.com/C-CINA/TomographyTools TomographyTools]. Once you created the ''.defocus'' file, you can continue in IMOD and click the button “correct CTF”.

=== Tomogram thickness ===
When defining the thickness of the tomogram consider [[Considerations_for_tomogram_positioning_in_IMOD | these points]].

=== Post processing in IMOD ===
In the post processing step in IMOD make sure that you swap the ''y'' and ''z'' dimensions and do not select “convert to bytes” in order to make the tomogram compatible with ''Dynamo''. Also make sure to keep all the IMOD files, in particular the ''.fid'' and ''.com'' files. This will be useful in cases where tomograms have to be reprocessed.

=== Assessing tomogram quality ===
After having computed the final tomogram, we recommend to visually check the tomogram quality as explained [[Tomogram quality assessment | here]].

=== NovaCTF ===
If [https://github.com/turonova/novaCTF novaCTF] is used for the tomogram reconstruction and CTF correction, the wrapper found in [https://github.com/C-CINA/TomographyTools TomographyTools] can be used.

@@ Line 28: / Line 28: @@
 === Tomogram generation ===
-We recommend to use back projected and full size tomograms (no binning) for processing. If the contrast in your tomogram is very low, it might be helpful to start with SIRT reconstructed tomograms to facilitate the particle detection and the first particle alignment iterations. SIRT tomograms have usually higher contrast but a lower resolution. We therefore always recommend to switch to the back projected tomograms towards the end of your subtomogram averaging, to make sure that you get the most out of your data. You can do this switch easily by cropping the particles from the back projected tomogram using the table from your last subtomogram averging iteration that was done using the SIRT tomogram. To save time during tomogram reconstruction, you can use a SIRT-like filter (equivalent to 10-50 SIRT iterations) in the normal back projection mode, instead of doing a proper SIRT reconstruction. This will lead to similar results but save a lot of time. To save even more time, you can directly create a 2 times binned SIRT tomogram (by binning the fine aligned tilt series first). You will not lose any high resolution data since you will switch back to the full size tomogram at a later step anyway. If you start with such binned tomograms, do not forget to adapt the coordinates in the table accordingly, when switching back to the full sized tomogram.
+We recommend to use back projected and full size tomograms (no binning) for processing. If the contrast in your tomogram is very low, it might be helpful to start with SIRT reconstructed tomograms to facilitate the particle detection and the first particle alignment iterations. SIRT tomograms have usually higher contrast but a lower resolution. We therefore always recommend to switch to the back projected tomograms towards the end of your subtomogram averaging, to make sure that you get the most out of your data. You can do this switch easily by cropping the particles from the back projected tomogram using the table from your last subtomogram averging iteration that was done using the SIRT tomogram. To save time during tomogram reconstruction, you can use a SIRT-like filter (equivalent to 10-80 SIRT iterations) in the normal back projection mode, instead of doing a proper SIRT reconstruction. This will lead to similar results but save a lot of time. To save even more time, you can directly create a 2 times binned SIRT tomogram (by binning the fine aligned tilt series first). You will not lose any high resolution data since you will switch back to the full size tomogram at a later step anyway. If you start with such binned tomograms, do not forget to adapt the coordinates in the table accordingly, when switching back to the full sized tomogram.
 === Post processing in IMOD ===

@@ Line 26: / Line 26: @@
 === Tomogram thickness ===
 When defining the thickness of the tomogram consider [[Considerations_for_tomogram_positioning_in_IMOD | these points]].
 === Post processing in IMOD ===

@@ Line 22: / Line 22: @@
 === CTF estimation ===
-The CTF is ideally estimated on the drift corrected stack (but not exposure filtered and not aligned). A matlab wrapper for [http://grigoriefflab.janelia.org/ctffind4 ctffind4] can be found in [https://github.com/C-CINA/TomographyTools TomographyTools]. If the CTF is estimated outside of the IMOD pipeline, the “Run Ctf Plotter” button from IMOD has to be skipped and the results from ctffind4 are used instead. The output from ctffind4 needs to be converted first into a file format that can be read by IMOD with the function found in [https://github.com/C-CINA/TomographyTools TomographyTools]. Once you created the ''.defocus'' file, you can continue in IMOD and click the button “correct CTF”.
+The CTF is ideally estimated on the drift corrected stack (but not exposure filtered and not aligned). A matlab wrapper for [http://grigoriefflab.janelia.org/ctffind4 ctffind4] can be found in [https://github.com/C-CINA/TomographyTools TomographyTools]. If the CTF is estimated outside of the IMOD pipeline, the “Run Ctf Plotter” button from IMOD has to be skipped and the results from ctffind4 are used instead. The output from ctffind4 needs to be converted first into a file format that can be read by IMOD with the function found in [https://github.com/C-CINA/TomographyTools TomographyTools]. Once you created the ''.defocus'' file, you can continue in IMOD and click the button “correct CTF”. Be aware of the different [[Angle_of_astigmatism_convention | conventions regarding the angle of astigmatism]].
 === Tomogram thickness ===

@@ Line 1: / Line 1: @@
 This article contains some practical suggestions and examples for the pre-processing of tomographic tilt series and the reconstruction of tomograms. Be aware that there exist many other working strategies and softwares in addition to the examples shown here.
-=== Naming convention ===
+=== Naming convention and organizing data ===
-When dealing with large and growing datasets, we highly recommend to strictly follow a consistent file naming convention. A possible convention is to use a number for the batch (a batch stands for one data acquisition session) and a number for the tilt series within this batch. E.g., the tilt series 21 of batch 1 would be named ''b001ts021.st''. The same goes for all other related file types such as ''b001ts021.rec'', ''b001ts021.fid'' and so on.
+When dealing with large and growing datasets, we highly recommend to strictly follow a consistent file naming convention. A proposed convention is to use a number for the batch (a batch stands for one data acquisition session) and a number for the tilt series within this batch. E.g., the tilt series 21 of batch 1 would be named ''b001ts021.st''. The same goes for all other related file types such as ''b001ts021.rec'', ''b001ts021.fid'' and so on. We also recommend to organize all your data in a comprehensive and systematic way from the very beginning. One possible way of organizing your data is shown in the figure below. See more details on [[Tips_for_management_of_tomographic_data_sets | tips for management of tomographic data sets]].
   [[File:data_organisation_dynamo.jpg|thumb|center|200px| Example of organizing your data.]]