IsoNet-cryoET / spIsoNet

Overcoming the preferred orientation problem in cryoEM with self-supervised deep-learning
https://www.biorxiv.org/content/10.1101/2024.04.11.588921v1
MIT License
17 stars 4 forks source link

Example output #12

Open sven-lange opened 4 months ago

sven-lange commented 4 months ago

Hi,

I was wondering if you could upload the expected output files for the tutorial - in my case, the training and prediction runs finish without errors (but there are some warnings i.e. "Applied workaround for CuDNN issue, install nvrtc.so"), however, the input and output maps look very similar/unchanged.

Many thanks in advance! Sven

procyontao commented 4 months ago

Hi,

I uploaded expected output for in this folder for both anisotropy and misalignment correction. https://ucla.box.com/s/aghai8n4k7wmfexw3ns4yncg50g2z9ir

Here are orthogonal slices comparations of the results in the supplementary figure. image

The half maps when viewed in chimera may not seems to change too much. You may want to see details after postprocessing.