iscc.sample.Resection ¶

Resection(tumor, compartment='primary')

A resected tumour specimen, cut into the samples the assays take.

Parameters:

Name	Type	Description	Default
`tumor`	`GenotypeTumor`	The grown (count-based) tumour standing in for the resected specimen.	required
`compartment`	`(primary, met, both)`	Which grid the specimen is taken from. `"primary"` (default) resects only the primary tumour; `"met"` resects only the metastatic deposit (errors if the tumour has no metastasis); `"both"` resects both grids (falls back to the primary alone when there is no metastasis). This sets the deme set that :meth:`bisect` partitions and that :meth:`dissociate` / :meth:`slice` default to when `region` is `None`.	`"primary"`

Examples:

>>> spec = Resection(tumor)                          # primary by default
>>> cut, remainder = spec.bisect(frac=0.42)          # in-plane cut
>>> cd = spec.dissociate(cut, max_cells=50000)       # sequencing sample (full depth)
>>> section = spec.slice(remainder, depth_frac=0.5)  # imaging section (thin slice)
>>> met = Resection(tumor, compartment="met").dissociate()   # the metastasis instead

Notes

Both samplers leave tumor.cell_data untouched — take as many cuts as you like without redefining what the tumour itself exports.

Methods:

Name	Description
`bisect`	In-plane cut: split the specimen into two disjoint 2-D parts.
`dissociate`	Dissociate a 2-D part into a per-cell table for the sequencing assays.
`slice`	Cut a thin histology / Visium section: a depth cut of a 2-D part.

bisect ¶

bisect(frac=0.5, axis='x')

In-plane cut: split the specimen into two disjoint 2-D parts.

Cuts each grid of the resected :attr:compartment at frac of its side along axis and returns the deme indices on each side, ready to pass to :meth:dissociate / :meth:slice (or make_cell_data(region=)). For compartment="both" each grid is cut at the same fraction and the parts are unioned.

Parameters:

Name	Type	Description	Default
`frac`	`float`	Fraction of the grid (along `axis`) that falls in the first part.	`0.5`
`axis`	`(x, y)`	Cut axis: `"x"` splits by column, `"y"` by row.	`"x"`

Returns:

Type	Description
`(list of int, list of int)`	`(part, remainder)` deme-index lists partitioning the compartment.

dissociate ¶

dissociate(region=None, max_cells=None, install=False)

Dissociate a 2-D part into a per-cell table for the sequencing assays.

Materialises the part at FULL column depth (all ~K cells/deme, subsampled to max_cells); the cells lose their spatial position, as in a real dissociation.

Parameters:

Name	Type	Description	Default
`region`	`iterable of int`	Deme indices of the part to dissociate (e.g. the first element of :meth:`bisect`). `None` dissociates the whole resected :attr:`compartment` (the primary by default).	`None`
`max_cells`	`int`	Cap on materialised cells (a representative subsample above it).	`None`
`install`	`bool`	Also make this sample the tumour's own `cell_data`. Off by default: a sample is not the tumour, and installing it would silently redefine what `to_anndata(tumor)` and `tumor.write_h5ad(...)` export. Turn it on when you want the tumour-level helpers that read `tumor.cell_data` (`clones_from_clades`, `to_anndata(tumor)`) to work on exactly these cells.	`False`

Returns:

Type	Description
`dict`	The per-cell `cell_data` tables (pass to `bulkDNA` / `scDNA` / `scRNA`).

slice ¶

slice(region=None, depth_frac=0.5, max_cells=None, install=False)

Cut a thin histology / Visium section: a depth cut of a 2-D part.

Keeps depth_frac of each deme's 3-D column (default half — "take away half"), leaving the 2-D field intact, so a spatial assay sees the whole structure at a realistic ~one-layer density.

Parameters:

Name	Type	Description	Default
`region`	`iterable of int`	Deme indices of the part to section (e.g. the remainder from :meth:`bisect`). `None` sections the whole resected :attr:`compartment` (the primary by default).	`None`
`depth_frac`	`float`	Fraction of each deme's column kept in the slice. Note this thins the same 3-D column that `Visium(section_frac=...)` thins, so the two compose multiplicatively: keep one of them at `1.0` (the tutorials slice here and leave `section_frac=1.0`).	`0.5`
`max_cells`	`int`	Cap on materialised cells (bounds the section at cm-scale).	`None`
`install`	`bool`	Also make this section the tumour's own `cell_data` — see :meth:`dissociate`. Off by default, so taking a section never changes what the tumour itself exports.	`False`

Returns:

Type	Description
`dict`	The per-cell `cell_data` tables of the thin section (pass to `Visium`).