Fixes #18

panpipes/funcs/plotting.py

@@ -35,28 +35,53 @@ def scatter_one(group_choice, col_choice, plot_df, axs=None, colour="#1f77b4", t
 
 
 def batch_scatter_two_var(plot_df, method, batch, palette_choice=None):
+    """Plots facetted umaps, with each group within the method and batch highlighted as foreground.
+    Args:
+        plot_df (pd.DataFrame): pandas dataframe contianin umap coordinates plus method and batch columns.
+        method (str): method column in plot_df (this will be plotted one category per column)
+        batch (str): batch column in plot_df (this will be plotted one category per row)
+        palette_choice (list): List of colors to plot each rows foreground. Defaults to None.
+    Returns:
+        fig, ax : matplotlib subplots figure
+    """
     plot_df = plot_df[['umap_1', 'umap_2', method, batch]]
     plot_df = plot_df.dropna()
     # n_vars = len(df[var_choice].unique())
-    nrows = len(plot_df[batch].unique())
-    ncols = len(plot_df[method].unique())
-    fig, axs = plt.subplots(nrows, ncols, figsize=(4*ncols, 3*nrows), facecolor='w', edgecolor='k')
-    fig.subplots_adjust(hspace=.2, wspace=.2)
-    if np.max([nrows, ncols]) > 1:
-        axs = axs.ravel(order="F")
+    method_choices = plot_df[method].cat.categories.tolist()
+    logging.debug(method_choices)
+    group_choices = plot_df[batch].unique()
+    logging.debug(group_choices)
+    nrows = len(group_choices)
+    ncols = len(method_choices)
+    if nrows > 40:
+        logging.info("skipping facet plot as too many variables: %s" % batch)
+        fig = None
+        axs = None
     else:
-        axs=[axs]
-    if palette_choice is None:
-        palette_choice = ['#1f77b4']*ncols
-    # for j in range(ncols):
-    idx=0
-    for j, method_choice in enumerate(plot_df[method].cat.categories.tolist()):
-        plot_df2 = plot_df[plot_df[method] == method_choice].copy()
-        for i in range(nrows):
-            group_choice = plot_df2[batch].unique()[i]
-            logging.debug(str(method_choice) + "|" + str(group_choice))
-            scatter_one(group_choice, batch, plot_df2, axs[idx], colour=palette_choice[j], title=str(method_choice) + "|" + str(group_choice))
-            idx+=1
+        fig, axs = plt.subplots(nrows, ncols, figsize=(4*ncols, 3*nrows), facecolor='w', edgecolor='k')
+        fig.subplots_adjust(hspace=.2, wspace=.2)
+        if np.max([nrows, ncols]) > 1:
+            logging.debug('ravelling the matploltib as there are multipl columns')
+            axs = axs.ravel(order="F")
+        else:
+            axs=[axs]
+        if palette_choice is None:
+            palette_choice = ['#1f77b4']*nrows
+            logging.debug(len(palette_choice))
+        # for j in range(ncols):
+        idx=0
+        for i, method_choice in enumerate(method_choices):
+            plot_df2 = plot_df[plot_df[method] == method_choice].copy()
+            for j, group_choice in enumerate(group_choices):
+                logging.debug('plotting %i %s' % (j, group_choice))
+                logging.debug(str(method_choice) + "|" + str(group_choice))
+                scatter_one(group_choice, batch, plot_df2, axs[idx], 
+                            colour=palette_choice[i], 
+                            title=str(method_choice) + "|" + str(group_choice))
+                idx+=1
+                logging.info(idx)
+    return fig, axs
+
 
 def facet_scatter(x, y, c, **kwargs):
     """Draw scatterplot with point colors from a faceted DataFrame columns."""

panpipes/python_scripts/plot_umaps_batch_correct.py

@@ -16,7 +16,7 @@
 import sys
 import logging
 L = logging.getLogger()
-L.setLevel(logging.WARNING)
+L.setLevel(logging.INFO)
 log_handler = logging.StreamHandler(sys.stdout)
 formatter = logging.Formatter('%(asctime)s: %(levelname)s - %(message)s')
 log_handler.setFormatter(formatter)
@@ -116,8 +116,9 @@
         g = (g.map(sns.scatterplot, "umap_1", "umap_2", col, s=pointsize, linewidth=0))
         g.add_legend() 
         g.savefig(os.path.join(args.fig_dir, mod, "umap_method_" + str(col) + ".png"))
-        batch_scatter_two_var(plt_df, "method", col, palette_choice=palette_choice)
-        plt.savefig(os.path.join(args.fig_dir,mod,  "umap_method_facet_" + str(col) + ".png"), dpi=300)
+        fig, ax = batch_scatter_two_var(plt_df, "method", col, palette_choice=palette_choice)
+        if fig is not None:
+            fig.savefig(os.path.join(args.fig_dir, mod,  "umap_method_facet_" + str(col) + ".png"), dpi=300)
         plt.clf()
 
     ncats  = len(plt_df['method'].unique())