Imported .mrc files and found they have been rotated and mirror-imaged


Hi everyone,

I'm trying to use Signature to select the particles from my micrograph, given a template.
I would like to get coordinates the picked particle and angle respect to the template to calculate nearest neighbor distance and angle distribution. After importing, I find that both micrograph and template are rotated by counterclockwise 90 degree and mirror-imaged across the mirror plane lying along y = x of the rotated image.

I guess that this manipulation during the import might not affect the distance distribution; I'm not certain of how it affect the angle distribution, compared if doing the same thing with un-rotated and un-mirror imaged micrograph and template.

Can anyone suggest the way to counter this problem? Is it possible to adjust the program such that it doesn't do such things for the import?


Hi Champ,

My guess is that this simply has to do with the way in which Signature displays the image. Different programs (especially in EM) tend to start displaying pixels from top right and across, or bottom left and across, etc... This specific rotation seems a bit weird, but probably should not affect you angular and distance ditributions. I definitely would not worry about it, unless you are working with tilt pairs (which signature isn't really meant for, TiltPicker is a much better program in that case). What is most likely is that if you window out the particles, you will find that whatever program you used to display the mrc files which you used as a reference to determine that your micrographs were rotated displays the particles in the same orientation as they would be on the micrographs (i.e. not rotated or mirrored). Of course you probably want to check all of this.

In short, mirroring will not affect you angular distribution, as long as all of your micrographs are mirrored in the same way.
This has to do with how the program choose to display pixels, and could be changed, though I doubt that James (the program's author) will do so. You can always use the mrc suite of programs to apply compensatory rotations and mirror operations, though this may do more harm than good (if such things can really cause harm).

Hope this helped,


In reply to by Axel

Hi Axel,

Thank you for your answer and confirmation. I also figured out that it won't affect the angle distribution since the smallest relative angle will be the same. I don't really understand the tilt pairs, but I'll look into TiltPicker. Actually, my micrograph is 2D-AFM image and I use EMAN to change .tiff to .mrc of the micrograph and determine the average of the boxed particles. So that I import them to Signature and am trying to get it to select and choose the particles and export their coordinates and orientation. So far, I trust the Signature program to some extent. If you (or anyone) have more suggestion for a better way to do that, please feel free to help me.

Thanks again,

Hi Champ,

I don't really know much about AFM, but in that case, you may want to be extra careful that your final images are faithful with your original micrographs. In EM, our images are projections, so mirror operations don't really affect the data much at all, however, if I understand AFM, then your images are surface representations, which are qualitatively different, so that a mirror operation matters (if you are working with chiral molecules). In that case, you may find yourself having to mirror either your final images, or your reference.