spapros.ev.ProbesetEvaluator.plot_knn_overlap

ProbesetEvaluator.plot_knn_overlap(set_ids=None, selections_info=None, **kwargs)

Plot mean knn overlap over k

Parameters:

• set_ids (Optional[List[str]]) – List of probeset IDs. Check out ProbesetEvaluator.summary_results for available sets.

• selections_info (Optional[DataFrame]) –

  Information on selections for plotting. The dataframe includes:

  • selection ids or alternative names as index

  • column: path (mandatory if data=None): path to results csv of each selection (which contains number of clusters (as index) and one column containing the data to plot)

  • optional columns:

    • color: matplotlib color

    • linewidth: matplotlib linewidth

    • linestyle: matplotlib linestyle

    • <groupby>: some annotation that can be used to group the legend

  Note that the legend order will follow the row order in selections_info.

• **kwargs – Any keyword argument from knn_overlap().

Return type:

None

Example

import spapros as sp
adata = sp.ut.get_processed_pbmc_data()
selections = sp.se.select_reference_probesets(
    adata, methods=["PCA", "DE", "HVG", "random"], n=30, seeds=[0, 777], verbosity=0)
evaluator = sp.ev.ProbesetEvaluator(
    adata, verbosity=0, results_dir=None, scheme="custom", metrics=["knn_overlap"])
for set_id, df in selections.items():
    gene_set = df[df["selection"]].index.to_list()
    evaluator.evaluate_probeset(gene_set, set_id=set_id)

evaluator.plot_knn_overlap()