The thin filament-like structure data processing

Hi all,

I’m working on a filement-like protein structure. The protein seem like a filament, and by using the filament tracer I can successfully pick particles from the micrographs.

The question is the protein feature from 2D classification seem very heterogeneity, many filament seem curved,even I use small box size (192) extract particles, the problem still exist.

I would like to get more straight filament from the 2D classification,any suggestion for improvment?