Update code/process_input.py

code/process_input.py

@@ -5,11 +5,13 @@ def clean_data(input_set):
     try:
         if ',' in input_set:
             input_set = [i.strip() for i in input_set.split(',')]
+            initial_len = len(input_set)
             for i in input_set:
                 data = data.append(pd.Series([j.strip() for j in i.split('-')]), ignore_index=True)
             data.columns = ['uniprotID', 'wt', 'pos', 'mut']
         elif '\t' in input_set:
             input_set = [i.strip() for i in input_set.split('\t')]
+            initial_len = len(input_set)
             for i in input_set:
                 data = data.append(pd.Series([j.strip() for j in i.split('-')]), ignore_index=True)
             data.columns = ['uniprotID', 'wt', 'pos', 'mut']
@@ -36,6 +38,8 @@ def clean_data(input_set):
             data.at[i, 'datapoint'] = data.at[i, 'uniprotID'] + data.at[i, 'wt'] + str(data.at[i, 'pos']) + data.at[i, 'mut']
 
         data = data.astype(str)
+        if initial_len != len(data):
+            print(f'{initial_len- len(data)} of {initial_len} datapoints is omitted. Check your input.')
         return data
     except ValueError:
         st.write('Your input is in the wrong format. Please see the example.')