spapros.se.ProbesetSelector.plot_histogram

ProbesetSelector.plot_histogram(x_axis_keys=None, selections=None, penalty_keys=None, unapplied_penalty_keys=None, background_key=True, **kwargs)

Plot histograms of (basic) selections under given penalties.

The full selection procedure consists of steps partially based on basic score based selection procedures. This is an interactive plotting function to investigate if the constructed penalty kernels are well chosen.

Note

The green line shows a linear interpolation of the penalty scores which is only an approximation of the penalty function.

Parameters:

  • x_axis_keys (Optional[Dict[str, str]]) – Dictionary with the values from penaly_keys as keys and the column name of adata.var that was used to calcuate the respective penalty factors.

  • selections (Optional[List[str]]) –

    Plot the histograms of selections based on

    • ’pca’ : pca loadings based selection of prior genes

    • ’DE’ : genes selected based on differential expressed which are used as the ‘forest_DE_baseline’

    • ’marker’ : genes from the marker list

  • penalty_keys (Optional[Dict]) –

    Dictionary with the selection method name as key and the name of the column for adata.var containing the penalty factors to plot. Selections that are not in penalty_keys are plottet without penalty. If None (default), we use the columns defined in the selector for each selection. The respective attributes of the selector are:

    • ’pca’: pca_penalties

    • ’DE’: DE_penalties

    • ’marker’: m_penalties_adata_celltypes

  • unapplied_penalty_keys (Optional[Dict]) – Same as penalty_keys but for penalties that were not applied to the selection.

  • background_key (Union[bool, str]) – Key in adata.var for preselected genes (typically ‘highly_variable_genes’) to plot as background histogram . If True (default), g_key is used. If None no background is plottet. If ‘all’, all genes are used as background.

  • kwargs – Further arguments for selection_histogram().

Returns:

Figure can be showed (default True) and stored to path (default None). Change this with show and save in kwargs.

Return type:

None