univi.preprocessing.RNAPreprocessor

univi.preprocessing.RNAPreprocessor#

class univi.preprocessing.RNAPreprocessor(n_hvg=2000, *, layer='counts', target_sum=10000.0, scale=False, normalize_on_selected=True, clip=None, exclude_genes=(), min_cells=1)[source]#

Bases: object

RNA counts -> log1p normalized features -> optional train-fitted Z scores.

Seurat v3 HVGs are selected on raw training counts. normalize_on_selected=True normalizes library size over the selected HVGs (the convention used in the CITE-seq and Multiome analyses of the article); False normalizes over the complete reference feature panel before subsetting. n_hvg=None keeps all features. No fallback HVG method is chosen silently.

__init__(n_hvg=2000, *, layer='counts', target_sum=10000.0, scale=False, normalize_on_selected=True, clip=None, exclude_genes=(), min_cells=1)[source]#

Methods

__init__([n_hvg, layer, target_sum, scale, ...])

fit(adata)

fit_transform(adata)

transform(adata)

Contents