======================================
Flow plots with adjunct histograms
======================================

.. plot::

    # Read in data
    data = pd.read_csv('data/data_flow_with_adjunct_hist/data_flow_with_adjunct_hist.csv')
    
    # Randomly sample 10^4 samples from each condition to make representative flow diagram
    numSamples = 10**4
    small_data = data.groupby(['cond']).sample(n=numSamples, random_state=1)

    # General plotting params
    x = 'CFSE-A'
    y = 'sulfo-Cyanine5-A'
    hue = 'cond'
    cond_list = ['Ctrl', 'NIL', 'NILDD', 'NILDDRR']
    colormap = {'Ctrl': 'grey', 'NIL': 'crimson',
                'NILDD': 'purple', 'NILDDRR': 'dodgerblue'}

    # definitions for the axes
    left, width = 0.1, 0.65
    bottom, height = 0.1, 0.65
    spacing = 0.005

    rect_scatter = [left, bottom, width, height]
    rect_histx = [left, bottom + height + spacing, width, 0.2]
    rect_histy = [left + width + spacing, bottom, 0.2, height]

    # Set up figure
    fig = plt.figure(figsize=(6, 6))
    ax_scatter = plt.axes(rect_scatter)
    ax_scatter.tick_params(direction='in', top=True, right=True)
    ax_histx = plt.axes(rect_histx)
    ax_histx.set_axis_off()
    ax_histy = plt.axes(rect_histy)
    ax_histy.set_axis_off()

    # Set limits
    xlim = (10, 1*10**5)
    ylim = (5*10**1, 1*10**5)
    ax_scatter.set_xlim(xlim)
    ax_scatter.set_ylim(ylim)
    ax_histx.set_xlim(xlim)
    ax_histy.set_ylim(ylim)

    # Make density plots
    sns.kdeplot(ax=ax_scatter, data=small_data, x=x, y=y, hue=hue, hue_order=cond_list,
                log_scale=True, common_norm=False, 
                palette=colormap, alpha=0.7, fill=False, legend=True)

    # Plot histograms
    sns.kdeplot(ax=ax_histx, data=small_data, x=x, hue=hue, hue_order=cond_list,
                log_scale=True, common_norm=False, 
                palette=colormap, alpha=0.1, fill=True, legend=False)
    sns.kdeplot(ax=ax_histy, data=small_data, y=y, hue=hue, hue_order=cond_list,
                log_scale=True, common_norm=False, 
                palette=colormap, alpha=0.1, fill=True, legend=False)

    # Add in CFSE and EU lines
    CFSE_low_thresh = np.percentile(data[(data['cond'] == 'Ctrl')]['CFSE-A'],  20)
    EU_thresh = np.percentile(
        data[(data['cond'] == 'NIL') & (data['CFSE_cat'] == 'CFSE-low')]['sulfo-Cyanine5-A'], 50)
    ax_scatter.axvline(CFSE_low_thresh, 0, 1, color='black')
    ax_scatter.axhline(EU_thresh, 0, 1, color='black')

    # Title
    fig.suptitle('14 dpi')
    # Misc plotting stuff
    fig.tight_layout()  # Helps improve white spacing
    plt.show()