How to merge two rows in pandas dataframe and save its indexes in a new column

I want to merge rows in my input df_unique IF the list from one_one_3first column is the same as in zero_zero_3first AND inversely too (zero_zero_3first the same as one_one_3first) --> like the 0 and 1 row in the input df.

After merging, I want to receive a list of indexes of merged rows in a new column and update the genes_count column with the sum for merged rows.

To do that, I've created columns one_zero and zero_one to be able to group rows under desired conditions:

# create columns to be able to group rows
df_unique['one_zero'] = df_unique['one_one_3first'] + df_unique['zero_zero_3first']
df_unique['zero_one'] = df_unique['zero_zero_3first'] + df_unique['one_one_3first']

Here is my input df_unique with created columns one_zero and zero_oneto group rows:

                  one_one_3first             zero_zero_3first  genes_count                                                one_zero                                                zero_one
0    ['P1-12', 'P1-25', 'P1-28']  ['P1-22', 'P1-89', 'P1-92']           16  ['P1-12', 'P1-25', 'P1-28']['P1-22', 'P1-89', 'P1-92']  ['P1-22', 'P1-89', 'P1-92']['P1-12', 'P1-25', 'P1-28']
1    ['P1-22', 'P1-89', 'P1-92']  ['P1-12', 'P1-25', 'P1-28']           22  ['P1-22', 'P1-89', 'P1-92']['P1-12', 'P1-25', 'P1-28']  ['P1-12', 'P1-25', 'P1-28']['P1-22', 'P1-89', 'P1-92']
2     ['P1-26', 'P1-6', 'P1-92']  ['P1-12', 'P1-25', 'P1-28']            3   ['P1-26', 'P1-6', 'P1-92']['P1-12', 'P1-25', 'P1-28']   ['P1-12', 'P1-25', 'P1-28']['P1-26', 'P1-6', 'P1-92']
3    ['P1-12', 'P1-26', 'P1-89']  ['P1-25', 'P1-88', 'P1-90']            4  ['P1-12', 'P1-26', 'P1-89']['P1-25', 'P1-88', 'P1-90']  ['P1-25', 'P1-88', 'P1-90']['P1-12', 'P1-26', 'P1-89']

I performed grouping rows under desired conditions but the last three lines with calculating the sum in genes_count column don't work correctly (the order of output records is different than in output and genes count in the updated column for non_merged rows, e.g. 1 and 2, is incorrect):

# group rows
group = df_unique[['one_zero', 'zero_one']].apply(frozenset, axis=1)
df_unique_final = df_unique.groupby(group, as_index=False).first()

# ?try to update genes_count column with the sum for grouped rows?
genes_count_in_df_unique_final = df_unique.groupby(group, as_index=False, sort=False).agg({'genes_count': 'sum'})
df_unique_final = df_unique_final.drop(columns=['genes_count', 'one_zero', 'zero_one'])
df_unique_final['genes_count'] = genes_count_in_df_unique_final['genes_count']

and for that moment the output looks like that:

                  one_one_3first             zero_zero_3first  genes_count
0    ['P1-12', 'P1-25', 'P1-28']  ['P1-22', 'P1-89', 'P1-92']           38
1    ['P1-12', 'P1-90', 'P1-95']  ['P1-26', 'P1-88', 'P1-92']            3
2     ['P1-22', 'P1-6', 'P1-92']  ['P1-28', 'P1-88', 'P1-90']            4

So, my questions are:

1. what should I change to keep the same order of records in the output as in the input to perform the column genes_count with correct values for every row?

and

2. how to save the indexes of grouped rows in a new column?

to receive the final output like that:

                  one_one_3first             zero_zero_3first  genes_count    idxs_list
0    ['P1-12', 'P1-25', 'P1-28']  ['P1-22', 'P1-89', 'P1-92']           38          0,1
1     ['P1-26', 'P1-6', 'P1-92']  ['P1-12', 'P1-25', 'P1-28']            3            2
2    ['P1-12', 'P1-26', 'P1-89']  ['P1-25', 'P1-88', 'P1-90']            4            3

? Be grateful for any advice!

To keep the indexes, the easiest is just transforming it to a regular column, then do whatever you want.

df = df.reset_index(drop=False).rename(columns={'index': "original_index"})

Then, when you do all the changes you need, you can simply sort by that column with df.sort_values("original_index")

I dealt with that in this way:

# create column 'df_unique_index' to be able to keep row indexes after merging
df_unique = df_unique.reset_index(drop=False).rename(columns={'index': "df_unique_index"})

# merge records with the same 1/1 and 0/0 samples crosswise with keeping indexes of merged rows and counting the sum of genes
group = df_unique[['one_zero', 'zero_one']].apply(frozenset, axis=1)

df_unique_final_1 = df_unique.groupby(group).agg({'df_unique_index': (lambda x: list(x)), 'one_one_3first': 'first',
                                                  'zero_zero_3first': 'first','genes_count': 'sum'}).reset_index()

df_unique_final_1 = df_unique_final_1.drop(columns=['index']).rename(columns={'one_one_3first': 'one_one',
                                                                              'zero_zero_3first': 'zero_zero'}).\
    reindex(columns=['one_one', 'zero_zero', 'df_unique_index', 'genes_count'])

and obtained the desired result:

                   one_one              zero_zero df_unique_index  genes_count
0    [P1-12, P1-25, P1-28]  [P1-22, P1-89, P1-92]          [0, 1]           38
1    [P1-12, P1-90, P1-95]  [P1-26, P1-88, P1-92]           [538]            1
2     [P1-22, P1-6, P1-92]  [P1-28, P1-88, P1-90]      [539, 812]            9