moscot.utils.tagged_array.TaggedArray¶

class moscot.utils.tagged_array.TaggedArray(data_src, data_tgt=None, tag=Tag.POINT_CLOUD, cost=None)[source]¶

Interface to interpret array-like data for moscot.solvers.

It is used to extract array-like data stored in AnnData and interpret it as either cost matrix, kernel matrix or a point cloud, depending on the tag.

Parameters:

data_src (Union[ndarray[tuple[int, ...], dtype[floating]], Array]) – Source data.
data_tgt (Union[ndarray[tuple[int, ...], dtype[floating]], Array, None]) – Target data.
tag (Tag) – How to interpret data_src and data_tgt.
cost (Union[str, Callable[..., Any], None]) – Cost function when tag = 'point_cloud'.

Methods

from_adata(adata, dist_key, attr[, tag, ...])

Create tagged array from AnnData.

Attributes

`cost`	Cost function when `tag = 'point_cloud'`.
`data_tgt`	Target data.
`is_cost_matrix`	Whether `data_src` is a cost matrix.
`is_graph`	Whether `data_src` is a graph.
`is_kernel`	Whether `data_src` is a kernel matrix.
`is_point_cloud`	Whether `data_src` (and optionally) `data_tgt` is a point cloud.
`shape`	Shape of the cost matrix.
`tag`	How to interpret `data_src` and `data_tgt`.
`data_src`	Source data.