structboost.SimulationResult¶

class structboost.SimulationResult(counts, stage_labels, stage_sizes, marker_mask, gene_means, size_factors, batch_labels, params, gene_level=None, leaf_paths=None, hierarchy=None)[source]¶

Bases: object

Simulated counts together with the ground truth used to generate them.

Parameters:

counts (ndarray[tuple[Any, ...], dtype[int32]])
stage_labels (ndarray[tuple[Any, ...], dtype[int64]])
stage_sizes (ndarray[tuple[Any, ...], dtype[int64]])
marker_mask (numpy.ndarray)
gene_means (ndarray[tuple[Any, ...], dtype[float64]])
size_factors (ndarray[tuple[Any, ...], dtype[float64]])
batch_labels (ndarray[tuple[Any, ...], dtype[int64]])
params (dict[str, int | float | bool | str])
gene_level (ndarray[tuple[Any, ...], dtype[int64]])
leaf_paths (ndarray[tuple[Any, ...], dtype[int64]])
hierarchy (tuple[int, ...] | None)

counts¶

Integer UMI counts, shape (n, n_genes).

Type:: NDArray[np.int32]

stage_labels¶

Stage index per cell, shape (n,). -1 marks leftover background cells that belong to no stage (possible when n is not divisible by stageno).

Type:: NDArray[np.intp]

stage_sizes¶

Number of cells per stage, shape (stageno,).

Type:: NDArray[np.intp]

gene_level¶

Shape (n_genes,). Which level of the cell-type hierarchy each gene marks — 0 is the broadest — or -1 for a noise gene. With hierarchy=None every marker is level 0.

Type:: NDArray[np.intp]

leaf_paths¶

Shape (stageno, n_levels). Ancestry of each leaf population, so leaf_paths[k, d] is the node index of population k at depth d. These become the per-level cell labels in the AnnData wrapper.

Type:: NDArray[np.intp]

hierarchy¶

The branching factors used, or None for the flat layout.

Type:: tuple[int, …] | None

marker_mask¶

Boolean marker indicator, shape (stageno, n_genes). marker_mask[k, j] is True when gene j is a marker of stage k. Genes in the overlap between two consecutive stages are True in both rows.

Type:: BoolArray

gene_means¶

Baseline expression mean per gene (lambda_j), shape (n_genes,).

Type:: NDArray[np.float64]

size_factors¶

Per-cell library size factor, shape (n,). All exactly 1.0 when lib_size_sd == 0.

Type:: NDArray[np.float64]

batch_labels¶

Batch index per cell, shape (n,).

Type:: NDArray[np.intp]

params¶

Scalar simulation parameters. Contains no None values so it can be stored in adata.uns and round-tripped through h5ad.

Type:: dict[str, int | float | bool | str]

Methods

__init__(counts, stage_labels, stage_sizes, ...)

Attributes

`gene_level`
`hierarchy`
`leaf_paths`
`marker_genes`	Genes that are a marker of at least one stage, shape (n_genes,).
`counts`
`stage_labels`
`stage_sizes`
`marker_mask`
`gene_means`
`size_factors`
`batch_labels`
`params`

property marker_genes: numpy.ndarray¶: Genes that are a marker of at least one stage, shape (n_genes,).