scib.metrics.hvg_overlap

scib.metrics.hvg_overlap(adata_pre, adata_post, batch_key, n_hvg=500, verbose=False)

Highly variable gene overlap

Metric that computes the average percentage of overlapping highly variable genes per batch pre post integration.

Parameters:

  • adata_pre – Unintegrated anndata object

  • adata_post – Integrated anndata object

  • batch_key – Batch variable in adata_post.obs

  • n_hvg – Number of hvgs to compute per batch

Returns:

Average percentage of overlapping highly variable genes

The score can only be computed on feature spaces. No preprocessing is needed, as the function will perform highly variable gene selection.

Example

# full feature output
scib.me.hvg_overlap(adata_unintegrated, adata, batch_key="batch")