Dear Sjors and others,
I was wondering what "probability-weighted angular assignment" in
relion_refine means? I searched email archive and literature for this,
but wasn't able to find.
The question is actually related to Dieter's question in archive
000157. I plotted data_model_pdf_orient_class_1 in _data.star and found
one dominant angular assignment. These particles even form an individual
class when I perform unsupervised classification.
In my knowledge this usually will cause a distortion in the final
reconstruction, but was not really the case. So I wonder if what
"probability-weighted angular assignment" does is to downplay the
contribution of these abundant particles?
If not, do you think removing some abundant particles according to
their _LogLikeliContribution will help the quality of the
reconstruction?
Best regards,
Qiyang
--
Qiyang JIANG,
Predoc fellow, SCHAFFITZEL team,
EMBL Grenoble.
|