Hi all,
I'm part of the DiMaio lab at the University of Washington. We are trying to develop a method to use atomic models to guide particle sorting. Currently, we are trying to feed density maps generated from atomic models to 3D classification in RELION3. We want to integrate models very finely within the typical processing pipeline.
I have been feeding two higher resolution maps (8A) to a single iteration of 3D classification, separating the classes into corresponding star files and then sending each class on to refinement. For a relatively clean dataset from a collaborator, what comes out from refinement is one class that looks reasonable, but another that looks oddly noisy - more like a half map than a full map. After 25 iterations, the particle distributions from classification that are used for refinement give two reasonable maps. In addition, when I feed two maps that are not generated by models but produced by Relion to a single iteration of 3D classification, I get two classes that are not remotely protein-like after refinement.
I have been using the --firstiter_cc flag for my generated maps, but I was wondering if there was anything else that Relion does in the first iteration of classification that might be causing this issue compared to the final iterations.
Thanks,
Gabriella Reggiano
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
|