cospar.pp.get_highly_variable_genes

cospar.pp.get_highly_variable_genes(adata, normalized_counts_per_cell=10000, min_counts=3, min_cells=3, min_gene_vscore_pctl=85)

Get highly variable genes.

We assume that data preprocessing are already done, like removing low quality cells. It first perform count normalization, then variable gene selection.

Parameters

adata : AnnData object

normalized_counts_per_cell : int, optional (default: 1000)

count matrix normalization

min_counts : int, optional (default: 3)

Minimum number of UMIs per cell to be considered for selecting highly variable genes.

min_cells : int, optional (default: 3)

Minimum number of cells per gene to be considered for selecting highly variable genes.

min_gene_vscore_pctl : int, optional (default: 85)

Gene expression variability threshold, in the unit of percentile, for selecting highly variable genes. Range: [0,100], with a higher number selecting more variable genes.

Returns

• Modify adata.var[‘highly_variable’].

• If ‘highly_variable’ existed before, save a copy at adata.obs[‘highly_variable_old’]