Reconstruction of outlier filtered particles

[Cookiemaster33]

Hi there.

I was able to run dragon-ai very successfully for ab-initio reconstructions.

I ran through the Jupyter notebook pipeline and produced a .pkl file to sort for outlier and non-outlier particles.

I am trying to use cryoDGRN to reconstruct a volume from the respective .pkl files, but was not able to make it work.

Any recommendations?

Best

Toby

(P.s.: FYI. I realized that the Jupyter notebook script contains some errors. For most of the "config commands" it should be e.g. "config['training']['use_gt_poses']" instead of "config['use_gt_poses']"

[michal-g]

Hey Toby, sorry for the late reply — for clarification, what version of cryoDRGN are you using for these analyses (e.g. by running cryodrgn --version)? DRGN-AI is compatible with the 2.3.0 version of cryoDRGN it was originally branched off from, and not the more recent 3.x.x cryoDRGN versions. The former can be installed e.g. using pip install git+https://github.com/ml-struct-bio/cryodrgn.git@2.3.0.

Let us know if this helps!
-Mike