moscot.utils.tagged_array.DistributionContainer¶
- class moscot.utils.tagged_array.DistributionContainer(xy, xx, a, b, conditions, cost_xy, cost_xx)[source]¶
Data container for OT problems involving more than two distributions.
TODO
- Parameters:
xy (
Union
[ndarray
[Any
,dtype
[floating
]],Array
,None
]) – Distribution living in a shared space.xx (
Union
[ndarray
[Any
,dtype
[floating
]],Array
,None
]) – Distribution living in an incomparable space.a (
Union
[ndarray
[Any
,dtype
[floating
]],Array
]) – Marginals when used as source distribution.b (
Union
[ndarray
[Any
,dtype
[floating
]],Array
]) – Marginals when used as target distribution.conditions (
Union
[ndarray
[Any
,dtype
[floating
]],Array
,None
]) – Conditions for the distributions.cost_xy (
Literal
['euclidean'
,'sq_euclidean'
,'cosine'
,'pnorm_p'
,'sq_pnorm'
,'geodesic'
]) – Cost function when in the shared space.cost_xx (
Literal
['euclidean'
,'sq_euclidean'
,'cosine'
,'pnorm_p'
,'sq_pnorm'
,'geodesic'
]) – Cost function in the incomparable space.
Methods
from_adata
(adata, a, b[, xy_attr, xy_key, ...])Create distribution container from
AnnData
.Attributes
Whether the distribution contains data corresponding to the condition.
Whether the distribution contains data corresponding to the linear term.
Whether the distribution contains data corresponding to the quadratic term.