Dear all,
I am analyzing a dataset collected from a 300kV microscope using Apoferritin grid. The pixel size is 0.96 A (not calibrated by the model fitting yet) and particle images were not downsampled when extracted from images. Box size was set to 256 pix.
Would anyone explain why the FSC plot did not drop to zero? Is this a problem to use the plot like it is, or can it be improved by using a larger box size? Many thanks for any comments!
Your resolution is very high, congratulations! You are hitting Nyquist (resolution limited by sampling), and may want to check out https://www.biorxiv.org/content/10.1101/2023.06.09.544325v1
Can you see features in the map corresponding to this res? At this resolution you should see holes in tyrosines, clear spherical waters, clearly defined density for backbone carbonyls
3 Likes
what is the detector writing movies as, .tiff or .eer ? If you have a Falcon 4/4i, you can up sample raw .eer movies by n=2. This will greatly increase box size and computational time.
1 Like
Many thanks for the comment olibclarke, and yes, the resolution is high enough to be able to see holes in tyrosines. Please refer to a snapshot of the map.
2 Likes
Thanks for the comment Mark, and it was .eer collected from a Falcon4i camera. I did not mean to achieve high resolution but to calibrate the pixel size, so I did not try to upsampling the data except for the motion correction process.
Hope it is going well @Jason.G
If I do have .eer format, myself and others here have always said, [paraphrasing] “it is an interesting experiment to upsample to 8K, *if you have the time and resources.”
I usually fully process in 4K and then re-assign those particles to micrographs that are imported with upsampling = 2 and 0.5-1.0e/frame - they also go through patch motion correction, and patch CTF est.
Also with .eer and the Falcon4i it is getting hundreds of frames per movie, not hard set at 20,40,60,80 as for .tiff. I have seen slightly different results here as well if you want to change dose/frame.
If using AFIS you can also include the metadata for each optics group (the .xml files) https://guide.cryosparc.com/processing-data/tutorials-and-case-studies/tutorial-epu-afis-beam-shift-import. Not sure if it helps much for calibrations.
2 Likes
Excellent guide!
Just to add, unless I’m working at very high mag (0.5 Angpix or smaller) I usually import with upsampling 2 as I’ve lost count of the number of samples I wasn’t expecting to hit physical Nyquist and did. At 0.5-0.8 Angpix it’s a toss-up, and depends greatly on what sort of sample I’ve collected whether I upsample or not. At 0.9 or larger, I basically always work in 8K (I’ve done 16K once, and even broke super-resolution Nyquist, but that was in RELION - which was fun in itself as RELION no longer supports 16K sampling and supported it spottily when it did - and some time before CryoSPARC had RBMC)…
1 Like