We are thrilled to introduce fully automated data processing for repeat-target cryo-EM.
Using new tools in CryoSPARC, it is now possible to use automated workflows to obtain results that match or exceed those from manual processing.
We developed and tested a generalized automation strategy on 21 GPCR datasets. The automation strategy, encapsulated in a Workflow JSON file, can be adopted by users and extended to other target classes.
Read the full story on bioRxiv.
Please see cryosparc.com/automated-workflows for workflow files and instructions!
12 Likes
Hello,
At the CCPEM Spring Symposium two weeks ago, @apunjani invited us to share repeat-target datasets to stress-test this automated workflow. I suppose this discussion is a good place for it? if not let me know and I will send elsewhere (to the [address redacted] email maybe?).
It is good timing, as the group I work with just released three datasets collected on the same protein:
The first two are of the same enzyme in different redox conditions, which activate or inactivate the enzyme. The third one is a point mutant of the enzyme that is inactive in the redox condition permissive for the activity of the WT. The first two datasets contain two distinct global conformations (it is in fact continuous motion between the two, but the particles could still be assigned into the closest of the two conformations), and the different redox conditions cause the two conformations to occur in very different ratios, which we show is the regulatory mechanism. The third dataset, of the mutant enzyme, only contains one of the two conformations.
Everything is explained in our preprint, with the image processing strategy detailed in SI Figures 4, 5 and 6.
This is much fewer repeat-target datasets than the examples in your impressive GPCR preprint, but the necessary symmetry expansion / 3D classification / local refinement for these datasets make them a bit more complex to process than for an asymmetric particle, so could maybe help make the repeat-target workflow more general. The classification masks are all in the corresponding EMDB entries.
If you want to use these datasets and have questions, don’t hesitate to email me (address in my forum profile), I will be happy to help!
4 Likes
Hi @Guillaume thank you very much for sharing!
These datasets are great test cases for automation, and indeed the added complexity of symmetry expansion and classification (and the continuous heterogeneity) will be challenging! We will review your preprint and the processing details.
Thank you 
1 Like