Job: Patch Motion Correction
Patch-based motion correction in cryoSPARC corrects for both global motion (stage drift) and local motion (beam-induced anisotropic sample deformation). The algorithm is GPU accelerated, usually taking less than 10s per movie. No prior knowledge about particle position is required.
Below are examples of plots produced by patch motion correction that illustrate the computed trajectories. The first set of plots show overall motion correction (an actual trajectory plot, followed by x- and y-motion plots as functions of frame number within the movie).
Example trajectories for overall motion (stage drift).
Next, an example of anisotropic deformation. Each red circle represents the center of one "patch" of the image, and the curves protruding from each circle represents the motion undergone by that part of the sample. Notice how the motion of nearby patches is correlated—adjacent patches have somewhat similar motion. The patch motion correction algorithm imposes smoothness constraints on the motion, to avoid the fit being corrupted by random noise in the micrograph.
Example trajectories for different "patches" of a micrograph (beam induced anisotropic deformation).
Input: Raw movies in .mrc, .tiff, .eer, or mrc.bz2 format.
Common Parameters:
    Only process this many movies: Selects the first n movies to process, instead of processing the entire set. Helpful when working with a set of movies for the first time, to better understand the data quality.
    Low-memory mode: When working with very large movies on GPUs with low memory (less than, say, 16 GB), it is possible to run out of GPU memory. If this happens, turning on low-memory mode will usually allow the processing to proceed. This option makes processing slightly slower, so it is turned off by default.
    Output F-crop factor: Reduces the size of the output micrographs by "Fourier-cropping" (discarding Fourier components above a certain frequency). This reduces the size of the micrographs produced by the algorithm (which can speed up downstream processing), but also discards some of the highest frequency information in the images. This is particularly useful when dealing with super-resolution movies. An F-crop factor of 1.0 corresponds to no Fourier cropping, 0.5 discards half of the Fourier components, and so on.
Output:
    Non-dose weighted and dose-weighted micrographs.
Common Next Steps: Patch-based CTF Estimation.
Other links:
Last modified 7mo ago
Copy link