sc_toolbox.tools.remove_outliers#

sc_toolbox.tools.remove_outliers(cords, eps=1, min_samples=2)[source]#

Remove outlying cells based on UMAP embeddings with DBScan (density based clustering).

Call as: sub.obs[“d_cluster”] = remove_outliers(sub.obsm[“X_umap”], min_samples = 10)

Parameters:

  • cords – adata UMAP coordinates, typically adata.obsm[“X_umap”]

  • eps (int) – Maximum distance between two clusters to still be considered neighbors

  • min_samples (int) – Minimum samples of a cluster

Return type:

Categorical

Returns:

Pandas Categorical of clusters