michelixtrace reference cropped

[Neville Bethel]

Hello,

First here is my output version information:

 GUI from package Emspring-0.86.1661

Spring environment loaded.

EMAN 2.1 alpha2 (CVS 2013/08/07 17:01:09)

Your EMAN2 is running on:  Ubuntu 18.04.1 LTS 4.15.0-34-generic x86_64

Your Python version is:  2.7.2

Michelixtrace has been very useful for my work.

I have been having issues with one type of fiber. This is one is thicker than the others, about 220 angstroms.

When I run michelixtrace, the output PDFs show a cropped reference about half the width of the original fiber,
and this seems to be presenting issues with the cross correlation analysis. This does not change if I increase the estimated fiber width.

I attached the input and parameter file and the resulting PDF.

Is there a parameter that I should be setting differently?

Best regards,

Neville Bethel

University of Washington

[Carsten Sachse]

Dear Neville,

Your cross correlation map does not show a high correlation trace, which is why MICHELIXTRACE does not seem to work well. 

Is your reference the same scale/pixel sizeI than the micrograph ? I assume this is correct. 

You seem to have significant protein grains in the background, which give rise to structured background and therefore less distinct peaks.

I would try to improve the correlation map by providing a cleaner reference. If you used a class average (from a small manually cropped/segmented data set) you should be able to improve the correlation peaks extending to visible traces in cc-map. Try to monitor that as it will give you more accurate traces.

Best wishes,

Carsten