Lighter plots, more robust tex-escaping.

Rasterizing might make pdfs lighter.

Escaping underscore for tex-compatibility is now idempotent.

Also added a missing tex-compatibility escaping.

libworkflows/libworkflows/libworkflows.py

@@ -87,9 +87,17 @@ def strip_split(text):
     return split(strip(text), "\t")
 
 
+# import inspect
 def texscape(text):
     """Escapes underscores to make a latex-compatible text."""
-    return sub("_", r"\_", text)
+    # https://stackoverflow.com/a/2654130/1878788
+    # currframe = inspect.currentframe()
+    # callframe = inspect.getouterframes(currframe)
+    # print(f"Escaping {text}\n(Called by {callframe})")
+    # print(f"Escaping {text}")
+    # return sub("_", r"\_", text)
+    # To avoid double escape:
+    return sub(r"([^\\])_", r"\1\_", text)
 
 
 def ensure_relative(path, basedir):
@@ -302,7 +310,7 @@ def save_plot(outfile,
               plot_func,
               *args,
               title=None, format=None,
-              tight=True, equal_axes=False, square=False,
+              tight=True, equal_axes=False, square=False, rasterize=False,
               **kwargs):
     """*format* is needed when using multiple pages output."""
     # https://stackoverflow.com/a/10154763/1878788
@@ -333,7 +341,7 @@ def save_plot(outfile,
     if format is None:
         plt.savefig(outfile, **save_kwds)
     else:
-        plt.savefig(outfile, format=format, **save_kwds)
+        plt.savefig(outfile, format=format, rasterize=rasterize, **save_kwds)
     plt.close(plt.gcf())
 
 

small_RNA-seq/small_RNA-seq.snakefile

@@ -108,6 +108,8 @@ from sklearn import preprocessing
 from sklearn.decomposition import PCA
 import matplotlib as mpl
 # To be able to run the script without a defined $DISPLAY
+# https://github.com/mwaskom/seaborn/issues/1262
+#mpl.use("agg")
 mpl.use("PDF")
 #mpl.rcParams["figure.figsize"] = 2, 4
 mpl.rcParams["font.sans-serif"] = [
@@ -1767,6 +1769,8 @@ rule make_fold_heatmap:
             gene_colours.name = f"{wildcards.small_type}_{wildcards.fold_type}"
         #all_folds.drop(["cosmid", "name", "small_type"], axis=1)
         #all_folds[all_folds.columns.difference(["cosmid", "name", "small_type"])]
+        # https://github.com/mwaskom/seaborn/issues/1262
+        #mpl.use("agg")
         try:
             #labels_dict = merge_with(tuple, small_type2colour, DefaultCounter(all_folds.small_type))
             # https://stackoverflow.com/a/47396625/1878788
@@ -1776,10 +1780,11 @@ rule make_fold_heatmap:
             save_plot(
                 output.fold_heatmap, plot_fold_heatmap,
                 all_folds.drop(["cosmid", "name", "small_type"], axis=1),
-                gene_colours, labels_dict, tight=False)
+                gene_colours, labels_dict, tight=False, rasterize=True)
         except ValueError as e:
             print(labels_dict)
             raise
+        #mpl.use("PDF")
 
 
 #def plot_norm_counts(counts_data, summaries):
@@ -3348,6 +3353,9 @@ def plot_text(outfile, text, title=None):
     # https://stackoverflow.com/a/21833883/1878788
     txt.set_clip_on(False)
     if title is not None:
+        usetex = mpl.rcParams.get("text.usetex", False)
+        if usetex:
+            title = texscape(title)
         plt.title(title)
     plt.tight_layout()
     plt.savefig(outfile)
@@ -3363,8 +3371,8 @@ def plot_ACGT(data, xlabel, ylabel):
     letter2legend = dict(zip("ACGTN", "ACGTN"))
     usetex = mpl.rcParams.get("text.usetex", False)
     if usetex:
-        xlabel = sub("_", r"\_", xlabel)
-        ylabel = sub("_", r"\_", ylabel)
+        xlabel = texscape(xlabel)
+        ylabel = texscape(ylabel)
         data.columns = [texscape(colname) for colname in data.columns]
     for (read_len, base_props) in data.iterrows():
         x_shift = -0.25