Curate exposure - any ML methods for selecting micrographs?

Hi all:

Is there a job in cryoSPARC, or some machine learning models that can take a small subset of manually accepted/ rejected micrographs together with their ctf, ice thickness, defocus etc. values and learn from that to filter a large dataset of micrographs?

Thanks!
Guanqi

Hi Guanqi

There is nothing like that in cryoSPARC, but you might want to check out MicAssess from Mike Cianfrocco:

https://www.sciencedirect.com/science/article/pii/S0969212620300800

Cheers
Oli

Not exactly, but GitHub - IsoNet-cryoET/spIsoNet: Overcoming the preferred orientation problem in cryoEM with self-supervised deep-learning was good per-particle.

In cryosparc you can perhaps set custom workflows https://guide.cryosparc.com/application-guide-v4.0+/workflows from pre-defined parameters of micrographs to run the whole SPA pipeline and see how the final volumes look.

e.g. compare thick ice with thin ice, defocus range (max = -3.0 vs. -2.0, etc).