package API about squidpy HOT 2 CLOSED

submodules:

• graph: graph always in obsp -> API is functional on anndata

• image: image features in obsm (maybe later in layers? once different number of features allowed). we keep all traditional stuff in an anndata instance, put image related things into an image sized xarray and give both instances to our methods that require both. we need to try to use dask to handle this xarray in a "backed" mode, ie reading from disk.

  • ImageContainer: class wrapping xarray, supporting lazy loading/cropping. Has crop(self, x, y, sizef, scalef) function
  • iter_crops(adata, image_container, sizef, scalef, channels=None): image crop iterator
  • segment_image(image_container): calculate segmented image - optionally add to image_container
  • create_features_table(adata, crops_iter, features=[]): iterate over crops (iterator), calculate features and add to obsm in adata

• plotting: plot high res for images, e.g. segmentation, via napari, plotting of graph to check graph coarsity etc via networkx?

from squidpy.

this has been pretty much addressed

from squidpy.