This just changes __init__ to use anndata as input, which allows removing a lot of input validation code.
I also added minimal sparse data support for computing autocorrelations, but make dense once genes are selected for get local correlations. Using the correct format (csc for cells by genes) makes it just as fast as fully dense.
This just changes
__init__
to use anndata as input, which allows removing a lot of input validation code.I also added minimal sparse data support for computing autocorrelations, but make dense once genes are selected for get local correlations. Using the correct format (csc for cells by genes) makes it just as fast as fully dense.
Fixes #5
Also has the readthedocs merged in.