fix: added ESM to fig4 violin #46

playground.py

@@ -1,9 +1,21 @@
+# # %%
+# from src.data_processing.init_dataset import create_datasets
+
+# create_datasets(
+#     data_opt=['davis'],
+#     feat_opt=['foldseek'],
+#     edge_opt=['binary']
+# )
+# # %%
+# from src.utils.loader import Loader
+# d2 = Loader.load_dataset(data='davis', pro_feature='nomsa',
+#                         edge_opt='anm')
 # %%
-from src.data_processing.init_dataset import create_datasets
+from src.data_analysis.figures import fig4_pro_feat_violin, prepare_df
+
+df = prepare_df('results/model_media/model_stats.csv')
 
-create_datasets(
-    data_opt=['davis'],
-    feat_opt=['foldseek'],
-    edge_opt=['binary']
-)
-# %%
+#%%
+fig4_pro_feat_violin(df, sel_dataset='davis', sel_col='mse',
+                     add_stats=True, verbose=False)
+# %%

src/data_analysis/figures.py

@@ -269,10 +269,11 @@ def fig4_pro_feat_violin(df, sel_dataset='davis', verbose=False, sel_col='cindex
 
     # Annotation for stats
     if add_stats:
-        pairs=[('nomsa', 'msa'), ('nomsa', 'shannon'), ('msa', 'shannon')]
+        pairs=[('nomsa', 'msa'), ('nomsa', 'shannon'), ('msa', 'shannon'),
+               ('ESM', 'nomsa'), ('ESM', 'msa'), ('ESM', 'shannon')]
         annotator = Annotator(ax, pairs, data=filtered_df, x='feat', y=sel_col, verbose=verbose)
         annotator.configure(test='Mann-Whitney', text_format='star', loc='inside', 
-                            hide_non_significant=True)
+                            hide_non_significant=not verbose)
         annotator.apply_and_annotate()
 
     if show: