crested.pl.qc.sort_and_filter_cutoff

crested.pl.qc.sort_and_filter_cutoff#

crested.pl.qc.sort_and_filter_cutoff(adata, cutoffs=None, method='proportion', min_scores=None, model_name=None, max_k=2000, cmap='tab20', legend=True, class_labels=True, ax=None, plot_kws=None, cutoff_line_kws=None, score_line_kws=None, **kwargs)#

Plot the effect of different top_k cutoffs in sort_and_filter_regions_on_specificity(), which takes the top k regions per class.

This function plots the gini scores for those regions per class, indicating the specificity. If the specificity is too low for the lower-ranked genes, decrease top_k.

Parameters:

  • adata (AnnData) – AnnData object with region data. This should not be filtered yet!

  • cutoffs (int | list[int] | None (default: None)) – (List of) considered top # of regions to plot an illustrative line at, as in sort_and_filter_regions_on_specificity()’s top_k.

  • model_name (str | None (default: None)) – The name of the model to calculate scores from. If None or ‘truth’/’groundtruth’/’X’ (default), will use the values in adata.X.

  • method (Literal[‘gini’, ‘proportion’] (default: 'proportion')) – The method to use for calculating scores, either ‘gini’ or ‘proportion’. Default is ‘proportion’.

  • min_scores (float | None (default: None)) – Minimum score(s) required to plot an illustrative line at, as in sort_and_filter_regions_on_specificity()’s min_score.

  • max_k (int (default: 2000)) – The maximum number of top regions per cell type to plot. Should be higher than the cutoffs you’re considering.

  • cmap (str | Sequence (default: 'tab20')) – Colors or colormap to draw colors for the different cell types from. Any valid input to color_palette() works. Examples are a matplotlib or seaborn colormap name or a list of colors.

  • legend (bool (default: True)) – Whether to add a cell type legend.

  • class_labels (bool (default: True)) – Whether to add labels on the per-class lines to the right edge of the plot.

  • ax (plt.Axes | None (default: None)) – Axis to plot values on. If not supplied, creates a figure from scratch.

  • plot_kws (dict | None (default: None)) – Extra keyword arguments passed to scatter(). Defaults: 's': 4.

  • cutoff_line_kws (dict | None (default: None)) – Extra keyword arguments passed to axvline() to show cutoffs. Defaults: 'color': 'black', 'linestyle': '--'.

  • score_line_kws (dict | None (default: None)) – Extra keyword arguments passed to axhline() to show min_score. Defaults: 'color': 'darkgray', linestyle: '-.'.

  • width – Width of the newly created figure if ax=None. Default is 8.

  • height – Height of the newly created figure if ax=None. Default is 5.

  • kwargs – Additional arguments passed to render_plot() to control the final plot output. Please see render_plot() for details. Custom defaults for ?: xlabel='Rank', ylabel='Gini score', grid='x'.

See also

crested.pl.render_plot, crested.pp.sort_and_filter_regions_on_specificity

Return type:

(plt.Figure, plt.Axes) | None

Example

>>> crested.pl.qc.sort_and_filter_cutoff(
...     adata,
...     cutoffs=[2000, 3000, 4000],
...     min_scores=[0.2, 0.3],
...     max_k=5000
... )
../../_images/qc_sort_and_filter_cutoff.png