Batch mag_distortion_correct

Submitted by davboyer on Mon, 2017-02-27 14:26

Forums

Magdistortion

Hi:

I have tested mag_distortion_estimate and mag_distortion_correct on 16 aligned and summed micrographs that have nice ice rings. Both these programs seem to work quite well, thank you for writing the software and making in easily available!

I am now trying to incorporate the mag_distortion_correct to my data processing steps for my current data set (1066 dose-fractionated movies). I am wondering if it is possible to use a command line command to correct all my movie stacks without having to combine them into a single stack and run the program interactively. Ideally, I could just submit a job for a given movie stack with all the input parameters mag_distortion_correct needs and then script it to run automatically for all my movies.

Is this possible with the mag_distortion_correct that is currently available on the website?

Thanks,

David

timgrant Mon, 2017-02-27 14:34

Hi David,

This can be accomplished with some fairly simple scripting. E.g a script to run mag distortion where you just supply the input would look something like the script copied below. You would then save it as something like correct_movie.sh and run it by calling "./correct_movie.sh input.mrc output.mrc". Obviously you will have change the answers to match your inputs for mag distortion, the example below is applying a gain reference and resizing the movie.

Cheers,

Tim

------
#!/bin/bash
#
./mag_distortion_correct_openmp_8_18_15.exe << eof
$1
$2
45.9
1.021
1.00
yes
gain_ref.mrc
yes
3838
3710
eof

davboyer Mon, 2017-02-27 15:28

Thanks! I'll give it a go today.

I noticed in the other messages, your recommended procedure is using the unbinned super-resolution pixel size (in my case 0.52 A) for input files to minimize interpolation effects. This is followed immediately by Fourier binning by 2 before further processing.

Just to double check, I'm assuming the re-sample option in mag_distortion_correct will perform the Fourier binning by 2 as long as you give the appropriate new X and Y image dimensions (e.g., super-res input X-dim=7676 Y-dim=7420 goes down to X-dim=3838 Y-dim=3710)?

Lastly, the final pixel size should be the 2x binned pixel size (in my case 1.04) divided by the major axis scale factor from the "not stretch-only" parameters. Correct?

Thanks in advance,

David