spatial-microenvironment-subset

spatial-microenvironment-subset

When to use

Extract centers and nearby observations. Use spatial-domains for global regions and spatial-condition for condition comparisons.

Use from a step

from skills._sdk.notebook import load_skill, read_input, write_output
library = load_skill("spatial-microenvironment-subset")
data = read_input("input.h5ad")
data = library.subset(data, center_key="cell_type", center_values=["Tumor"], radius_native=50)
write_output(library.selection_table(data), "tables/results.csv")

Run examples/example_step.py through the step runner.

API

subset(adata, *, center_values: list[str], center_key: str | None=None, radius_native: float | None=None, radius_microns: float | None=None, microns_per_coordinate_unit: float | None=None, data_type: str | None=None, include_centers: bool=True, target_key: str | None=None, target_values: list[str] | None=None)

Return a new neighborhood AnnData; expression matrices remain unchanged.

:param adata: Labelled AnnData with spatial coordinates; X, layers and raw are sliced. :param center_values: Labels defining centers, required as in the CLI. :param center_key: Label column; None selects the first recognized label column. :param radius_native: Positive coordinate-unit radius; default None requires radius_microns. :param radius_microns: Positive micron radius, exclusive with radius_native. :param microns_per_coordinate_unit: Explicit positive scale; None uses platform metadata. :param data_type: Optional platform hint used to resolve units, as in the CLI. :param include_centers: True retains centers regardless of the target-label filter. :param target_key: Optional neighbor label column; None uses the center column. :param target_values: Optional allowed neighbor labels; None admits all labels. :returns: A new AnnData with role, distance and JSON diagnostics. :raises ValueError: Invalid radius, labels, units, coordinates or an empty selection.