Hi all,
I am looking for a way to deal with preferential orientations in cryosparc at the 3D level rather at 2D classification. We are working with a small protein complex and rebalancing in 2D seems to merge and discard particles from distinct classes while still keeping too many of the preferred orientation ones. So the final 3D distribution is still quite biased but with less data whereas 3D models feature significant streaking artifacts and, as of the new non-uniform refinement, a crumpled surface of all features. We’ve played reasonably with ab initio classes, 3D heterorefinement, etc. but the problem persists. We have a lot of data and the data is of very good quality (Titan Krios, energy filter, K3 DED, no carbon edges, low defocus, collected by pro’s), the 2D classes are of remarkable contrast and detail, so any suggestions how to curate particle distribution after refinement or minimize noise artifact for small proteins at the 3D refinement level will be very much appreciated.
Thanks in advance,
Petya