cellpin.dataset.scAnnDataset

cellpin.dataset.scAnnDataset#

class cellpin.dataset.scAnnDataset(adata, layer=None, gene_symbols=None, panel=None, batch_key=None)#

Bases: Dataset

scRNA-seq AnnData dataset wrapper. Returned by cellpin.pp.setup_data().

Outputs (per observation):

  • full_expr: full expression row (all genes, stable order)

  • panel_expr: full_expr restricted to panel genes

  • no_panel_expr: full_expr restricted to non-panel genes

  • local_l_mean: dataset-level mean of log-library size (1,)

  • local_l_var: dataset-level variance of log-library size (1,)

  • batch_index: integer batch label (only when batch_key is set)

Args:

adata: scRNA-seq AnnData object. layer: Expression layer to read. Must be raw counts. When None,

.X is used.

gene_symbols: var column name for alternative gene identifiers

(same semantics as in setup_data).

panel: Ordered list of panel gene names. Internally, panel gene order

is fixed to ascending position in adata (boolean-mask order), so panel_expr[k] always corresponds to genes in that order, not the order of the panel argument.

batch_key: obs column for integer batch labels. When None,

batch conditioning is off.

Methods table#

__init__(adata[, layer, gene_symbols, ...])

Methods#