Well, I had a request to identify common and not-common elements between two frames in python (pandas). Not-common means not shared between either frames. In pandas, outer join terminology is confusing for SQL folks. Outer join means union in Pandas, in SQL, outer join means symmetric difference. Since no outer join (in terms of SQL) in pandas, we have a round about way. Let us go that way. 

Data frame1:

gene1 1.5183

gene2 1.49

gene3 1.4991

gene4 1.4104

gene5 0.964

gene6 59.19

gene7 118.4

Data frame 2:

gene1 1.5052

gene2 1.505

gene3 1.5037

gene8 1.7976

gene9 1.3445

gene10 117.68

gene6 60.3

gene7 116.99

Now user wants common elements between data frame 1 and data frame 2. This is easy in pandas. User also wants elements unique to each dataframe. This is tricky. 

Expected output of common elements:

gene1      1.5183    1.5052

gene2      1.49      1.505

gene3      1.4991    1.5037

gene6      59.19     60.3

gene7      118.4     116.99

Expected outcome of unique elements:

gene4           1.4104      –

gene5          0.964        –

gene8            –          1.7976

gene9            –          1.3445

gene10          –          117.68

Code:

----------------------

import pandas as pd

test1=pd.read_csv("test1.txt", sep="\t",  header=None)

test1.columns=["a","b"]

test2=pd.read_csv("test2.txt", sep="\t",  header=None)

test2.columns=["a","b"]

pd.merge(test1, test2, on='a', how='inner') # prints common elements

    a   b_x     b_y

0   gene1   1.5183  1.5052

1   gene2   1.4900  1.5050

2   gene3   1.4991  1.5037

3   gene6  59.1900     60.3000

4   gene7  118.4000    116.9900

outer_common=pd.merge(test1, test2, on='a', how='outer', indicator=True)

outer_common_nc=outer_common[outer_common['_merge']!="both"]

outer_common_nc.iloc[:,0:3] # prints unique elements.

a   b_x     b_y

3   gene4   1.4104  NaN

4   gene5  0.9640  NaN

7   gene8   NaN     1.7976

8   gene9  NaN     1.3445

9   gene10  NaN     117.6800

=============================

indicator in merge function gives the element presence in each data frame. First we created a frame with union (outer join in pandas). Then we filtered the data frame for elements that are not in both frames. After that we removed unwanted _merge column. Ofcourse, one can always replace NaN with whatever user wants.