Filament Tracing for large helical assemblies

Hello,

I would like hear if there are any suggestions for the processing of data shown below. I get some success with regular blob picking, but I am curious if filament tracing can be applied to this dataset. The helical diameter is about 200A, filament diameter is about 100A and a full turn is about 350A. Anything I have tried has not picked up the filaments at all, or will only pick up a subset of them (like a turn here and there). Any help appreciated!

Hi @boflynn,

Thanks for the post. This dataset looks interesting in that the filament appears to “zigzag” across the micrograph, rather than appearing as a straight line. The filament tracing was tuned mostly for straight filaments, so it might be a bit challenging to get it to perform better on this dataset.

For starters, check out our guide page on the filament tracer for an explanation of the parameters and diagnostic plots, as well as some suggestions to help tune the parameters. Some specific suggestions on this dataset:

  • You might have more success using template-free tracing, where the filament diameter is set to the diameter of the individual strand (I take it this is 100Å?), rather than diameter of the helix.
    This would hopefully give picks that are located on the “elbows” and straight regions of the filament. Ideally, the “elbow”-centered picks could then be discarded during inspect picks (likely high curvature?) and during 2D classification.
  • Alternatively, if you’re using pre-selected templates, it’s important that these cover all possible views along the helical axis. So ideally there should be views centered on the “elbow” of the filament, as well as on the straight portion.

If you like, I’d also be happy to take a look at the diagnostic plots – please DM me if so.

Best,
Michael

1 Like

Hi @mmclean

Thanks for your answer! I will DM you soon W.R.T. this dataset. As we screen different mutated forms of this protein we can see that these helix structures are ubiquitous so I would love to optimize how we analyze them.

Thanks,
Brian