Hi, I recently 3D-classified the membrane protein and validated the results with Heterogeneous reconstruction only. By my inspection, a Class showed the feature that I wanted, and all transmembrane helices looked great. But soon, however, after applying non-uniform or local refinements on the class, a transmembrane helix that was adjacent to a ‘feature’ gone bad, which was cut and never showed helix backbone. In this case, which method can give me reliable insights about the classification? It’s very confusing because I used reconstruction-only jobs very frequently…
Plus, the broken site is the boundary of the membrane micelle and solvent area…I hope this might be helpful..
Hi @Randomnick, thanks for your question!
We are interested to learn more about your processing pipeline and results to understand what might have happened, so it would be great if you could let us know:
- How do the class volumes look if you use hard classification in Hetero Reconstruction?
- What do the ESS values/plots look like from 3D Classification? If you could share the ESS histogram from 3D class that would be useful.
- How many particles were you classifying, into how many classes, and how many particles were in the class that had the features you were interested in?
- What does your consensus map look like (images would be great if you can share) and how were the poses estimated? Was the upstream job a single refinement?
- Did you apply a mask during 3D Classification? If so, how large was this, and are you able to show an image of the mask relative to the consensus volume?