==========
Histograms
==========

Function to make subplots for several variables
------------------------------------------------

.. plot::

    # ----- PLOTTING FUNCTION -------

    def plot_hists(df,              # DataFrame with data
                   x,               # variable (str, column header of df) or list of variables to plot
                   sample_list,     # list of samples (values of column 'ConditionLabel') to plot in order
                   name,            # short name (str) for saving the figure
                   plot_title=None, # Super-title for plot (optional str)
                   xlim=(1e1,1e6),  # tuple or list of tuples (optional) for x-axis range on subplot(s)
                   gates=None       # list (optional) of x-values at which to draw vertical lines, i.e. gates
                  ):

        # Specify additional parameters
        lin = set(['FSC-A','FSC-H','FSC-W','SSC-A','SSC-H','SSC-W']) # list of variables to plot on linear scale (size-related params)
        condition_palette = pd.read_pickle('data/histogram/exp42_palette.pkl') # custom color palette (dict mapping Condition values to colors)
        sns.set_context('talk',rc={'font.family': 'sans-serif', 'font.sans-serif':['Helvetica Neue']})

        sz = len(x) # number of subplots
        fig, axs = plt.subplots(1,sz,figsize=(sz*7,5))

        # Create each subplot
        for i in range(sz):
            g = sns.kdeplot(data=df, x=x[i], ax=axs[i],
                            hue='ConditionLabel', hue_order=sample_list,    # samples are specified by the 'ConditionLabel' column in df
                            palette=condition_palette,
                            fill=True, alpha = 0.1,                         # shade under the curve, but faintly
                            log_scale=(x[i] not in lin),                    # default to log scale unless variable is in the linear list
                            legend=(i==sz-1),                               # add a legend only to the last subplot
                            common_norm=True)                               # supposed to normalize area under the curve (?)

            # Formatting
            sns.despine(ax=g)      # remove top and right plot border
            g.minorticks_off()

            # Set the x-axis range based on provided list or value
            if isinstance(xlim,list): g.set_xlim(xlim[i])
            else: g.set_xlim(xlim)

            # Add a vertical line (gate) if specified
            if gates is not None: g.axvline(gates[i], c='gray', ls='--', alpha=0.5)

        # Adjust the legend to the right of the last subplot
        sns.move_legend(axs[-1], title='Condition', loc='upper left', bbox_to_anchor=(1,1), frameon=False)

        # Add a super-title to the plot, if specified
        fig.suptitle(plot_title)
        fig.tight_layout()

        # [NOT APPLICABLE HERE] Save the figure as an image to the path specified by output_path (not defined here)
        #   uses rushd outfile function to save metadata associated with the figure
        # fig.savefig(rd.outfile(output_path/('hist-'+name+'.svg')),bbox_inches='tight')

        return fig
    # ----- END PLOTTING FUNCTION -------

    # ----- Load data -----
    labeler = pd.read_pickle('data/histogram/exp42_labeler.pkl') # dict mapping short Condition name to long ConditionLabel
    data = pd.read_csv('data/histogram/exp42_data-small.csv') # stored DataFrame of data with metadata

    # ----- Call the plotting function in a loop to generate plots with subsets of samples -----

    # Specify variables -> this will generate two subplots
    x = ['mGL-A', 'mCherry-A']

    # Filter data
    df = data.loc[(data[x[0]]>0) & (data[x[1]]>0)] # remove cells with log-unfriendly values
    # Normalize number of cells in each sample (Condition) by downsampling to smallest sample size
    num_cells = df.groupby(['Condition','Replicate'])[x[0]].count().min()
    df = df.groupby(['Condition']).sample(n=num_cells, random_state=1)

    # Dictionary of plots (separate figures) to generate
    #  keys = short name that plots are saved under
    #  values = list of samples (values in 'Condition' column) to include in the plot
    plot_list = {
        'tandem': ['260_False','260_True','261_False','261_True','262_False','262_True'],
        'divergent+dox': ['263_True','264_True','265_True'],
        'hPGK': ['260_False','260_True','263_False','263_True'],
    }

    # Loop to create each plot/figure
    for name, sample_list in plot_list.items():

        # Convert short sample names above (Condition) to long names (ConditionLabel) -> this is particular to my df organization
        sample_list = [labeler[s] for s in sample_list]
        # Identify the subset of the data to plot
        dd = df.loc[df['ConditionLabel'].isin(sample_list)]

        plot_hists(dd, x, sample_list, name,
                   xlim=[(1e0,1e6),(2e0,1e7)],  # specify a different x-axis range for each variable
                   gates=[3e2,0])               # draw a gate on the first subplot but not the second