Pandas str.replace skipping some replacements

Question

I have a column ms_sample_id in a pandas dataframe named selected_id_df. The unique rows are given by

selected_id_df.ms_sample_id.unique()

Which yields:

   array(['mitra_baseline_310808-1', 'mitra_baseline_310808-2',
   'mitra_baseline_310808-3', 'mitra_baseline_310808-4',
   'mitra_baseline_310907-1', 'mitra_baseline_310907-2',
   'mitra_baseline_310907-3', 'mitra_baseline_310907-4',
   'mitra_baseline_311090-1', 'mitra_baseline_311090-2',
   'mitra_baseline_311090-3', 'mitra_baseline_311090-4',
   'mitra_baseline_311091-1', 'mitra_baseline_311091-2',
   'mitra_baseline_311091-3', 'mitra_baseline_311091-4',
   'mitra_baseline_311123-1', 'mitra_baseline_311123-2',
   'mitra_baseline_311123-3', 'mitra_baseline_311123-4',
   'frozen-2w_310808-1', 'frozen-2w_310808-2', 'frozen-2w_310907-1',
   'frozen-2w_310907-2', 'frozen-2w_311090-1', 'frozen-2w_311090-2',
   'frozen-2w_311091-1', 'frozen-2w_311091-2', 'frozen-2w_311123-1',
   'frozen-2w_311123-2', 'RT-2w_310808-1', 'RT-2w_310808-2',
   'RT-2w_310907-1', 'RT-2w_310907-2', 'RT-2w_311090-1',
   'RT-2w_311090-2', 'RT-2w_311091-1', 'RT-2w_311091-2',
   'RT-2w_311123-1', 'RT-2w_311123-2', 'LT_RT_310808_1',
   'LT_RT_310808_2', 'LT_RT_310907_1', 'LT_RT_310907_2',
   'LT_RT_311090_1', 'LT_RT_311090_2', 'LT_RT_311091_1',
   'LT_RT_311091_2', 'LT_RT_311123_1', 'LT_RT_311123_2',
   'LT-frozen_310808_1', 'LT-frozen_310808_2', 'LT-frozen_310907_1',
   'LT-frozen_310907_2', 'LT-frozen_311090_1', 'LT-frozen_311090_2',
   'LT-frozen_311091_1', 'LT-frozen_311091_2', 'LT-frozen_311123_1',
   'LT-frozen_311123_2'], dtype=object)

I would like to replace some of the - in the ids with _. I do this as follows:

selected_id_df.loc[:,'ms_sample_id'] = (selected_id_df.loc[:,'ms_sample_id'] 
                                   .str.strip()
                                   .str.replace("frozen_2w", 'frozen-2w')
                                   .str.replace("RT_2w", 'RT-2w')
                                   .str.replace('mitra_baseline', 'mitra-baseline')
                                   .str.replace('LT_RT', 'LT-RT')                                      
                                   .str.replace('-1', '_1')
                                   .str.replace('-2', '_2')
                                   .str.replace('-3', '_3')
                                   .str.replace('-4', '_4'))

After running the above statement, I again use

selected_id_df.ms_sample_id.unique()

Which this time yields:

array(['mitra-baseline_310808_1', 'mitra-baseline_310808_2',
   'mitra-baseline_310808_3', 'mitra-baseline_310808_4',
   'mitra-baseline_310907_1', 'mitra-baseline_310907_2',
   'mitra-baseline_310907_3', 'mitra-baseline_310907_4',
   'mitra-baseline_311090_1', 'mitra-baseline_311090_2',
   'mitra-baseline_311090_3', 'mitra-baseline_311090_4',
   'mitra-baseline_311091_1', 'mitra-baseline_311091_2',
   'mitra-baseline_311091_3', 'mitra-baseline_311091_4',
   'mitra-baseline_311123_1', 'mitra-baseline_311123_2',
   'mitra-baseline_311123_3', 'mitra-baseline_311123_4',
   'frozen_2w_310808_1', 'frozen_2w_310808_2', 'frozen_2w_310907_1',
   'frozen_2w_310907_2', 'frozen_2w_311090_1', 'frozen_2w_311090_2',
   'frozen_2w_311091_1', 'frozen_2w_311091_2', 'frozen_2w_311123_1',
   'frozen_2w_311123_2', 'RT_2w_310808_1', 'RT_2w_310808_2',
   'RT_2w_310907_1', 'RT_2w_310907_2', 'RT_2w_311090_1',
   'RT_2w_311090_2', 'RT_2w_311091_1', 'RT_2w_311091_2',
   'RT_2w_311123_1', 'RT_2w_311123_2', 'LT-RT_310808_1',
   'LT-RT_310808_2', 'LT-RT_310907_1', 'LT-RT_310907_2',
   'LT-RT_311090_1', 'LT-RT_311090_2', 'LT-RT_311091_1',
   'LT-RT_311091_2', 'LT-RT_311123_1', 'LT-RT_311123_2',
   'LT-frozen_310808_1', 'LT-frozen_310808_2', 'LT-frozen_310907_1',
   'LT-frozen_310907_2', 'LT-frozen_311090_1', 'LT-frozen_311090_2',
   'LT-frozen_311091_1', 'LT-frozen_311091_2', 'LT-frozen_311123_1',
   'LT-frozen_311123_2'], dtype=object)

​We can see that my replacement statement has not worked for 2 of my replacements:

.str.replace("frozen_2w", 'frozen-2w')
.str.replace("RT_2w", 'RT-2w')

I am incredibly puzzled, why is this happening?

Thank you

Answer 1

Try this regex solution:

s.replace('-(?=[0-9])','_', regex=True).replace('_(?=2w|baseline|RT)','-', regex=True)

Explanation:

• We use regex to find - before a number [0-9] and replace with _
• We use regex again to find _ before 2w or baseline or RT and replace with -

Exemplified:

import pandas as pd

s = pd.Series(['mitra_baseline_310808-1', 'mitra_baseline_310808-2',
   'mitra_baseline_310808-3', 'mitra_baseline_310808-4',
   'mitra_baseline_310907-1', 'mitra_baseline_310907-2',
   'mitra_baseline_310907-3', 'mitra_baseline_310907-4',
   'mitra_baseline_311090-1', 'mitra_baseline_311090-2',
   'mitra_baseline_311090-3', 'mitra_baseline_311090-4',
   'mitra_baseline_311091-1', 'mitra_baseline_311091-2',
   'mitra_baseline_311091-3', 'mitra_baseline_311091-4',
   'mitra_baseline_311123-1', 'mitra_baseline_311123-2',
   'mitra_baseline_311123-3', 'mitra_baseline_311123-4',
   'frozen_2w_310808-1', 'frozen-2w_310808-2', 'frozen-2w_310907-1',
   'frozen-2w_310907-2', 'frozen-2w_311090-1', 'frozen-2w_311090-2',
   'frozen-2w_311091-1', 'frozen-2w_311091-2', 'frozen-2w_311123-1',
   'frozen-2w_311123-2', 'RT-2w_310808-1', 'RT-2w_310808-2',
   'RT-2w_310907-1', 'RT-2w_310907-2', 'RT-2w_311090-1',
   'RT-2w_311090-2', 'RT-2w_311091-1', 'RT-2w_311091-2',
   'RT-2w_311123-1', 'RT-2w_311123-2', 'LT_RT_310808_1',
   'LT_RT_310808_2', 'LT_RT_310907_1', 'LT_RT_310907_2',
   'LT_RT_311090_1', 'LT_RT_311090_2', 'LT_RT_311091_1',
   'LT_RT_311091_2', 'LT_RT_311123_1', 'LT_RT_311123_2',
   'LT-frozen_310808_1', 'LT-frozen_310808_2', 'LT-frozen_310907_1',
   'LT-frozen_310907_2', 'LT-frozen_311090_1', 'LT-frozen_311090_2',
   'LT-frozen_311091_1', 'LT-frozen_311091_2', 'LT-frozen_311123_1',
   'LT-frozen_311123_2'])

print(s.replace('-(?=[0-9])','_', regex=True).replace('_(?=2w|baseline|RT)','-', regex=True))