Allen Cell Discussion Forum

Aicsshparam/cvapipe_analysis - questions about adding new alignment functions

In the cell variance paper, it is noted that the spherical harmonics coefficients are sensitive to the orientations of the cells, and that different methods of alignment can generate different versions of shape space. I am currently working on writing a function to align my cells in a way that is different from the default alignment_2d method, and I have a couple of questions:

  1. Are there plans to add new alignment functions, and would there be interest in adding functions contributed by the community? I imagine this package is mostly being used internally at the moment, but it could be a neat addition in the future to “crowdsource” alignment functions used for different types of data,

  2. I was also wondering the purpose of the make_unique option when aligning the cells, which wasn’t used for the paper. What would be a situation where one might want to use make_unique = True?

By the way, thanks for the update to the readme file for aicsshparam, the animation looks cool and the examples are helpful. :slight_smile:

Hi @lynn. We are working on different methods for alignment and investigating how they impact the final shape space. It would be great to know more about what you have in mind. Could you create an issue on the GitHub repo for us to continue the discussion there? That could lead to an eventual PR and merge.

Now about your question on the make_unique flag. When this is set to False, the same cell can end up being aligned either to clockwise or counter-clockwise depending on the initial orientation. If you set that to True, then alignment becomes unique, meaning the same cell is always aligned the same way regardless the original orientation. Does that make sense?

We are glad you like the new doc in the repo. See you on GitHub.

1 Like