Output st_data.obsm['COVET'] of scenvi.compute_covet has the shape nx64x64, irrespective of k and g

[KarolineHoller]

Dear authors,
I was running compute_covet on Xenium data for different values of the parameters k and g. According to your paper, obsm["COVET"] should be a tensor of the size n x k x g, with n being the number of cells, k the number of neighbors and g the number of HVG genes. However, the shape of this tensor is always n x 64 x 64. Is it possible that there is still a hard coded parameter somewhere in your code?

Can you please provide a minimal example of how to cluster the niches from the covariance matrix? Also a reproducibility notebook for your analysis for Xenium data would be highly appreciated.

Thank you!