Visualization

Soup correlation plot

Scatter plot comparing the aggregate cell expression profile vs the soup profile (log10 scale). Genes that fall far above the diagonal in cells are potential contamination markers.

from SoupX import plot_soup_correlation

fig = plot_soup_correlation(sc, save_path=None)

Marker distribution

Violin plot of observed/expected expression ratio for marker gene sets across clusters.

from SoupX import plot_marker_distribution

gene_list = {'HB': ['HBB', 'HBA2', 'HBA1']}
fig = plot_marker_distribution(sc, non_expressed_gene_list=gene_list)

Marker map

2D scatter plot of the dimension reduction coloured by expression of a marker gene.

from SoupX import plot_marker_map

# Requires sc.DR to be set (loaded automatically by load_10x when tSNE/UMAP present,
# or set manually via set_dr())
fig = plot_marker_map(sc, gene='HBB', dr='umap')

Change map

Before/after comparison of a gene's expression in the 2D embedding.

from SoupX import plot_change_map

corrected = adjust_counts(sc)
fig = plot_change_map(sc, corrected, gene='HBB', dr='umap')

Downstream embedding

After running run_downstream, plot the UMAP coloured by cluster labels:

from SoupX.downstream import plot_embedding, plot_top_de_genes

plot_embedding(
    embedding = result['embedding'],   # (n_cells × 2)
    labels    = result['cluster_labels'],
    title     = 'UMAP - corrected',
    save_path = None,
)

plot_top_de_genes(result['de_results'], top_n=5)

Saving figures

All plot functions accept an optional save_path argument:

fig = plot_soup_correlation(sc, save_path='figures/soup_correlation.png')

When save_path is set, the figure is saved at 150 DPI and closed automatically. When None, the figure is displayed interactively.