Hi all,
Cryo-EM newbie here.
After many rounds of 2D classification, I have some not completely terrible looking classes of a ~135 kDa homodimeric transcription factor. Resolution decreases after homo- and heterogeneous refinement, but the best looking volume docks quite well with the AF3 predicted structure so I don’t think this it totally garbage. What are ways to improve 2D classes and preserve map resolution in 3D reconstruction + downstream processing?
Processing parameters are as follows:
-
Ab initio from 60k ptcls with 2 classes, initial minibatch size 400, final minibatch size 1200
-
Hetero refine with forced hard classification
