Merge pull request #92 from jjc2718/figures_final

Larger figure fonts for publication

00_download_data/nbconverted/3_sample_intersection.py

@@ -33,14 +33,13 @@
 
 
 # map data types to readable names
-# TODO: store this somewhere central
 data_map = {
-    'expression': 'gene expression',
-    'me_27k': '27k methylation',
-    'me_450k': '450k methylation',
+    'expression': 'expression',
+    'me_27k': '27k',
+    'me_450k': '450k',
     'rppa': 'RPPA',
     'mirna': 'microRNA',
-    'mut_sigs': 'mut sigs',
+    'mut_sigs': 'mut. sigs',
 }
 
 # get sample list for each -omics data type
@@ -115,12 +114,11 @@ def series_from_samples(samples, labels):
 # In[8]:
 
 
-# probably stick with venn diagram here
 sns.set({'figure.figsize': (8, 10)})
 sns.set_style('white')
 
 # get only sample lists from gene expression and mutation
-labels = ['gene expression', 'mutation']
+labels = ['expression', 'mutation']
 label_map = {l: sample_lists[l] for l in labels}
 
 venn(label_map)
@@ -133,7 +131,7 @@ def series_from_samples(samples, labels):
 
 
 sns.set_style('white')
-labels = ['gene expression', 'mutation', '27k methylation', '450k methylation']
+labels = ['expression', 'mutation', '27k', '450k']
 samples = [sample_lists[l] for l in labels]
 
 upset_series = series_from_samples(samples, labels)
@@ -143,11 +141,13 @@ def series_from_samples(samples, labels):
 # In[10]:
 
 
-subplots = up.plot(upset_series[upset_series != 0], element_size=60)
-plt.title('TCGA sample intersections, expression/methylation datasets', size=13)
-plt.ylabel('Intersection size', size=13)
-plt.yticks(fontsize=13)
-subplots['matrix'].set_yticklabels(labels=subplots['matrix'].get_yticklabels(), fontsize=12)
+subplots = up.plot(upset_series[upset_series != 0], element_size=60,
+                   totals_plot_elements=1)
+plt.title('TCGA sample intersections, expression/methylation datasets', size=18)
+plt.ylabel('Intersection size', size=18)
+plt.yticks(fontsize=18)
+subplots['matrix'].set_yticklabels(labels=subplots['matrix'].get_yticklabels(), fontsize=18)
+plt.gcf().delaxes(subplots['totals'])
 
 # we can clean up whitespace below in figure assembly script
 if SAVE_FIGS:
@@ -164,8 +164,8 @@ def series_from_samples(samples, labels):
 
 
 sns.set_style('white')
-labels = ['gene expression', 'mutation', '27k methylation', '450k methylation',
-          'RPPA', 'microRNA', 'mut sigs']
+labels = ['expression', 'mutation', '27k', '450k',
+          'RPPA', 'microRNA', 'mut. sigs']
 samples = [sample_lists[l] for l in labels]
 
 upset_series = series_from_samples(samples, labels)
@@ -176,10 +176,11 @@ def series_from_samples(samples, labels):
 
 
 subplots = up.plot(upset_series[upset_series >= 100], element_size=60)
-plt.title('TCGA sample intersections, all datasets', size=13)
-plt.ylabel('Intersection size', size=13)
-plt.yticks(fontsize=13)
-subplots['matrix'].set_yticklabels(labels=subplots['matrix'].get_yticklabels(), fontsize=12)
+plt.title('TCGA sample intersections, all datasets', size=18)
+plt.ylabel('Intersection size', size=16)
+plt.yticks(fontsize=16)
+subplots['matrix'].set_yticklabels(labels=subplots['matrix'].get_yticklabels(), fontsize=16)
+plt.gcf().delaxes(subplots['totals'])
 
 if SAVE_FIGS:
     images_dir = Path(cfg.images_dirs['data'])

02_classify_mutations/nbconverted/plot_all_results.py

@@ -19,14 +19,6 @@
 import matplotlib.pyplot as plt
 import seaborn as sns
 
-# svgutils sometimes doesn't deal well with text as paths in svgs
-# so for this file, we'll save the text as text, then convert it by hand
-# in inkscape to paths which seem to work fine in svgutils
-# 
-# for some reason just the plots in this file have this issue, plots from
-# other analysis scripts don't have this issue so we don't have to do all this
-plt.rcParams['svg.fonttype'] = 'none'
-
 import mpmp.config as cfg
 import mpmp.utilities.analysis_utilities as au
 import mpmp.utilities.plot_utilities as plu
@@ -345,7 +337,7 @@
 if merged_geneset:
     # for the larger gene set we need a bigger figure, since we
     # separate the heatmap into multiple rows
-    sns.set({'figure.figsize': (38, 13)})
+    sns.set({'figure.figsize': (38, 21)})
 else:
     sns.set({'figure.figsize': (28, 6)})
 
@@ -360,7 +352,7 @@
                                 origin_eps_y=0.015,
                                 length_x=0.85,
                                 length_y=0.95)
-    plt.suptitle('Performance by data type for cancer-related genes, all data types')
+    plt.suptitle('Performance by data type for cancer-related genes, all data types', size=26)
     plt.tight_layout()
 else:
     ax = plu.plot_heatmap_split(heatmap_df,
@@ -427,10 +419,3 @@
     plt.savefig(images_dir / 'all_heatmap_sig.png',
                 dpi=300, bbox_inches='tight')
 
-
-# In[20]:
-
-
-print(heatmap_sig_df.shape)
-heatmap_sig_df.iloc[:, :10]
-

02_classify_mutations/nbconverted/plot_expression_gene_sets.py

@@ -149,7 +149,7 @@
 all_results_df.sort_values(by='p_value').head(10)
 
 
-# In[7]:
+# In[13]:
 
 
 sns.set({'figure.figsize': (24, 6)})
@@ -171,7 +171,7 @@
                           SIG_ALPHA,
                           metric=plot_metric,
                           verbose=True,
-                          label_x_lower_bounds=[0.2, 0.2, 0.4],
+                          label_x_lower_bounds=[0.2, 0.3, 0.5],
                           label_y_lower_bounds=[4, 4, 5],
                           mark_overlap=True,
                           overlap_reference='cancer gene set')

02_classify_mutations/nbconverted/plot_methylation_results.py

@@ -20,14 +20,6 @@
 import seaborn as sns
 from adjustText import adjust_text
 
-# svgutils sometimes doesn't deal well with text as paths in svgs
-# so for this file, we'll save the text as text, then convert it by hand
-# in inkscape to paths which seem to work fine in svgutils
-# 
-# for some reason just the plots in this file have this issue, plots from
-# other analysis scripts don't have this issue so we don't have to do all this
-plt.rcParams['svg.fonttype'] = 'none'
-
 import mpmp.config as cfg
 import mpmp.utilities.analysis_utilities as au
 import mpmp.utilities.plot_utilities as plu
@@ -148,7 +140,7 @@
                           SIG_ALPHA,
                           metric=plot_metric,
                           verbose=True,
-                          label_x_lower_bounds=[0.5, 0.4, 0.4],
+                          label_x_lower_bounds=[0.55, 0.4, 0.4],
                           label_y_lower_bounds=[4, 4, 4])
 
 if SAVE_FIGS:

05_classify_mutations_multimodal/nbconverted/plot_multimodal_results.py

@@ -31,7 +31,7 @@
 
 
 BAYES_OPT = False
-MLP = True
+MLP = False
 
 if BAYES_OPT:
     results_dir = Path(
@@ -226,7 +226,8 @@
     handle = mpatches.Patch(color=colors[ix], label=data)
     handles.append(handle)
 
-plt.legend(title='Data types used to train model', handles=handles, loc='lower right')
+plt.legend(title='', handles=handles,
+           loc='lower right', fontsize=15, title_fontsize=15)
 plt.tight_layout()
 
 if SAVE_FIGS:
@@ -408,7 +409,7 @@
 compare_df[compare_df.gene == 'TP53'].head(10)
 
 
-# In[ ]:
+# In[17]:
 
 
 sns.set({'figure.figsize': (13, 6)})
@@ -426,9 +427,3 @@
     plt.savefig(images_dir / svg_filename, bbox_inches='tight')
     plt.savefig(images_dir / png_filename, dpi=300, bbox_inches='tight')
 
-
-# In[ ]:
-
-
-
-