amici.interpretation.AMICIExplainedVarianceModule

amici.interpretation.AMICIExplainedVarianceModule#

class amici.interpretation.AMICIExplainedVarianceModule(_adata=None, _explained_variance_df=None, _compute_kwargs=None, _n_permutations=None)[source]#

Bases: object

Methods

__init__

compute

Creates a DataFrame with the explained variance scores.

compute_max_explained_variance_head

Compute the head with the maximum explained variance for the given cell type.

plot_explained_variance_barplot

Plots a barplot of the maximum explained variance score for each head and each cell type.

plot_featurewise_explained_variance_heatmap

Plots a heatmap of the explained variance scores for each head for n_top_genes genes.

save

Save explained variance scores to file
classmethod compute(model, adata=None, alpha=0.05, run_permutation_test=True, n_permutations=None)[source]#

Creates a DataFrame with the explained variance scores.

DataFrame contains the explained variance scores for each head across cells for all cell types or a subset of cell types.

Parameters:

  • adata (AnnData, optional) – The AnnData object.

  • alpha (float, optional) – The desired significance level for the explained variance of each head-cell type pair via permutation testing. Defaults to 0.05.

  • run_permutation_test (bool, optional) – Whether to run the permutation test. Defaults to True.

  • n_permutations (int, optional) – The number of permutations to use for the permutation test. If not provided, the minimum number of permutations required to achieve a significance level of alpha will be used.

Return type:

AMICIExplainedVarianceModule

Returns:

AMICIExplainedVarianceModule: The module instance with computed explained variance scores stored in _explained_variance_df with columns:

  • head: The index of the head.

  • ct_name: The name of the cell type.

  • gene: The name of the gene.

  • expl_var_head_gene: The explained variance score when all but head is ablated for the gene.

  • expl_var_head: The explained variance score when all but head aggregated across all genes.

if run_permutation_test is True, the following columns are also added:

  • p_value_head: The p-value of the explained variance score when all but head is ablated.

  • p_value_adj_head: The adjusted p-value of the explained variance score when all but head is ablated.

save(save_path)[source]#

Save explained variance scores to file

compute_max_explained_variance_head(cell_type)[source]#

Compute the head with the maximum explained variance for the given cell type.

Parameters:

cell_type (str) – The cell type to compute the maximum explained variance head for.

Return type:

int

Returns:

int: The head with the maximum explained variance for the given cell type.

plot_explained_variance_barplot(palette=None, cell_type_sub=None, wandb_log=False, show=True, save_png=False, save_svg=False, save_dir='./figures')[source]#

Plots a barplot of the maximum explained variance score for each head and each cell type.

Parameters:

  • expl_variance_df (pd.DataFrame) – A DataFrame containing the explained variance scores for each head and each cell type as returned by model.get_expl_variance_scores().

  • palette (str, optional) – The color palette to use for the barplot. Defaults to None.

  • cell_type_sub (list[str], optional) – A list of cell types to include in the barplot. Defaults to None.

  • wandb_log (bool, optional) – Whether to log the plot to Weights and Biases. Defaults to True.

  • show (bool, optional) – Whether to display the plot. Defaults to False.

  • save_png (bool, optional) – Whether to save the plot as a PNG file. Defaults to False.

  • save_svg (bool, optional) – Whether to save the plot as an SVG file. Defaults to False.

  • save_dir (str, optional) – The directory to save the plot files. Defaults to “./figures”.

plot_featurewise_explained_variance_heatmap(cell_type_sub=None, n_top_genes=20, wandb_log=False, show=True, save_png=False, save_svg=False, save_dir='./figures')[source]#

Plots a heatmap of the explained variance scores for each head for n_top_genes genes.

Parameters:

  • expl_variance_df (pd.DataFrame) – A DataFrame containing the explained variance scores for each head and each cell type as returned by `model.get_expl_variance_scores()”

  • cell_type_sub (list[str], optional) – A list of cell types for which to plot the heatmap.

  • n_top_genes (int, optional) – The number of top genes to include in the heatmap.

  • wandb_log (bool, optional) – Whether to log the plot to Weights and Biases.

  • show (bool, optional) – Whether to display the plot.

  • save_png (bool, optional) – Whether to save the plot as a PNG file.

  • save_svg (bool, optional) – Whether to save the plot as an SVG file.

  • save_dir (str, optional) – The directory to save the plot files. Defaults to “./figures”. Saves the filename as “expl_variance_heatmap_{cell_type}.png” or “expl_variance_heatmap_{cell_type}.svg”.