Update app.py

app.py

@@ -28,7 +28,7 @@ import requests
 from rdkit.DataStructs import BulkTanimotoSimilarity
 from requests.adapters import HTTPAdapter, Retry
 from markdown import markdown
-from rdkit import Chem, DataStructs
+from rdkit import Chem
 from rdkit.Chem import AllChem, Draw, RDConfig, PandasTools, Descriptors, rdMolDescriptors, rdmolops, Lipinski, Crippen
 from rdkit.Chem.Scaffolds import MurckoScaffold
 import seaborn as sns
@@ -833,7 +833,8 @@ def submit_predict(predict_filepath, task, preset, target_family, opts, job_info
             predictions, _ = predict(cfg)
             predictions = pd.concat([pd.DataFrame(prediction) for prediction in predictions], ignore_index=True)
             predictions['Source'] = f'Predicted ({preset} {target_family})'
-            prediction_df = pd.concat([prediction_df, predictions])
+            df_list = [prediction_df, predictions]
+            prediction_df = pd.concat([df for df in df_list if not df.empty])
 
         else:
             predictions_file = f'{SERVER_DATA_DIR}/{job_id}_{task_file_abbr[task]}_family-recommended_predictions.csv'
@@ -885,10 +886,12 @@ def submit_predict(predict_filepath, task, preset, target_family, opts, job_info
                 predictions = pd.concat([pd.DataFrame(prediction) for prediction in predictions], ignore_index=True)
                 predictions['Source'] = (f'Predicted ({row["Model"]} '
                                          f'{family.title() if row["Type"] == "Family" else "General"})')
-                prediction_df = pd.concat([prediction_df, predictions])
+                df_list = [prediction_df, predictions]
+                prediction_df = pd.concat([df for df in df_list if not df.empty])
 
         prediction_df = prediction_df.merge(orig_df, on=['X1', 'X2'], how='left', indicator=False)
-        prediction_df = pd.concat([prediction_df, annotated_df], ignore_index=True)
+        df_list = [prediction_df, annotated_df]
+        prediction_df = pd.concat([df for df in df_list if not df.empty], ignore_index=True)
 
         if "Include Compound Max. Tanimoto Similarity to Training Compounds" in opts:
             for family in prediction_df['Target Family'].unique():
@@ -1020,6 +1023,7 @@ def update_df(file, progress=gr.Progress(track_tqdm=True)):
 
 def create_html_report(df, file=None, task=None, opts=(), progress=gr.Progress(track_tqdm=True)):
     df_html = df.copy(deep=True)
+    df_html.dropna(how='all', axis=1, inplace=True)
     column_aliases = COLUMN_ALIASES.copy()
     cols_left = list(pd.Index(
         ['ID1', 'X1', 'Scaffold SMILES', 'Compound', 'Scaffold', 'ID2', 'X2', 'Y^']).intersection(df_html.columns))
@@ -1110,27 +1114,31 @@ def create_html_report(df, file=None, task=None, opts=(), progress=gr.Progress(t
         styled_df = df_html.fillna('').style.format(precision=3)
 
         for i, col in enumerate(num_cols):
+            cmap = sns.light_palette(num_col_colors[i], as_cmap=True)
             if col in df_html.columns:
                 if col not in ['Binding Affinity (IC50 [nM])']:
+                    cmap.set_bad('white')
                     styled_df = styled_df.background_gradient(
-                        subset=[col], cmap=sns.light_palette(num_col_colors[i], as_cmap=True))
+                        subset=[col], cmap=cmap)
                 else:
+                    cmap = cmap.reversed()
+                    cmap.set_bad('white')
                     styled_df = styled_df.background_gradient(
-                        subset=[col], cmap=sns.light_palette(num_col_colors[i], as_cmap=True).reversed())
+                        subset=[col], cmap=cmap)
 
         if any(df_html.columns.isin(bool_cols)):
-            styled_df.applymap(lambda val: f'background-color: {bool_col_colors[val]}', subset=bool_cols)
+            styled_df.map(lambda val: f'background-color: {bool_col_colors[val]}', subset=bool_cols)
 
-        table_html = styled_df.to_html(na_rep='')
+        table_html = styled_df.to_html()
         unique_html = ''
         if unique_df is not None:
             if 'Target FASTA' in unique_df.columns:
                 unique_df['Target FASTA'] = unique_df['Target FASTA'].str.replace('\n', '<br>')
             if any(unique_df.columns.isin(bool_cols)):
-                unique_df = unique_df.style.applymap(
+                unique_df = unique_df.style.map(
                     lambda val: f"background-color: {bool_col_colors[val]}", subset=bool_cols)
             unique_html = (f'<div style="font-family: Courier !important;">'
-                           f'{unique_df.to_html(escape=False, index=False, na_rep="")}</div>')
+                           f'{unique_df.to_html(escape=False, index=False)}</div>')
 
         return (f'<div style="font-size: 16px; font-weight: bold;">{job} Report Preview (Top 30 Records)</div>'
                 f'<div style="overflow-x:auto; font-family: Courier !important;">{unique_html}</div>'
@@ -1176,15 +1184,19 @@ def create_html_report(df, file=None, task=None, opts=(), progress=gr.Progress(t
             disabled=True, sizing_mode='stretch_both', pagination='local', page_size=30)
 
         for i, col in enumerate(num_cols):
+            cmap = sns.light_palette(num_col_colors[i], as_cmap=True)
             if col not in ['Binding Affinity (IC50 [nM])']:
                 if col not in ['Interaction Probability']:
+                    cmap.set_bad(color='white')
                     report_table.style.background_gradient(
-                        subset=df_html.columns == col, cmap=sns.light_palette(num_col_colors[i], as_cmap=True))
+                        subset=df_html.columns == col, cmap=cmap)
                 else:
                     continue
             else:
+                cmap = cmap.reversed()
+                cmap.set_bad(color='white')
                 report_table.style.background_gradient(
-                    subset=df_html.columns == col, cmap=sns.light_palette(num_col_colors[i], as_cmap=True).reversed())
+                    subset=df_html.columns == col, cmap=cmap)
 
         pie_charts = {}
         for y in df_html.columns.intersection(['Interaction Probability', 'Binding Affinity (IC50 [nM])']):