比较两列并获得熊猫的唯一值

Question

I have a dataframe in which some values are in two different columns

Ligand_hit,Ligand_miss
M00001,M00005
M00002,M00001
M00003,M00007
M00004,M00003

I would like to create a new column with all values in "Ligand_miss" that are not in "Ligand_hit". The desired output would be something like:

Ligand_hit,Ligand_miss,Unique
M00001,M00005,M00005
M00002,M00001,M00007
M00003,M00007,NaN
M00004,M00003,NaN

I tried to use "pandas.isin", but it only outputs boolean values. Is there a simple way to get the desired results?

Answer 1

A direct pandas solution can be this one:

df["Unique"] = df["Ligand_miss"][~df["Ligand_miss"].isin(df["Ligand_hit"])].drop_duplicates()

  Ligand_hit Ligand_miss  Unique
0     M00001      M00005  M00005
1     M00002      M00001     NaN
2     M00003      M00007  M00007
3     M00004      M00003     NaN

this provides index-lookup for the unique values.

Answer 2

You can simply use vanilla python, thanks to set :

In [129]: df
Out[129]: 
  Ligand_hit Ligand_miss
0     M00001      M00005
1     M00002      M00001
2     M00003      M00007
3     M00004      M00003

In [130]: pd.concat([df, pd.Series(list(set(df['Ligand_miss'].values) - set(df['Ligand_hit'].values)))], ignore_index=True, axis=1)
Out[130]: 
        0       1       2
0  M00001  M00005  M00007
1  M00002  M00001  M00005
2  M00003  M00007     NaN
3  M00004  M00003     NaN

Some explanations:

• set(df['Ligand_miss'].values) and set(df['Ligand_hit'].values) get the unique values in the 2 columns.

• set(...) - set(...) computes the difference (the "Unique") per your requirements.

• pd.concat merges the result into the original dataframe.

Answer 3

A basic list comprehension will do:

[i for i in df.Ligand_miss if i not in df.Ligand_hit]

You can also use sets for this:

list(set(df.Ligand_miss)-set(df.Ligand_hit))

Answer 4

There is a function in Pandas called isin() . You can use that to find the values from Ligand_miss which are in Ligand_hit . The reverse of which is the values from Ligand_miss which are not in Ligand_hit . Then you have to subset your data frame based on the reverse and save it in a new column. For example:

Lets say you have the data frame items_data as below:

 col_a col_b a_1 b_1 a_2 b_2 a_3 a_3 a_4 b_4 a_5 b_5 

You can create a new column called col_def by this line of code:

 items_data['col_def'] = items_data['col_a'][~items_data['col_a'].isin(items_data['col_b'])] 

This will give you items from column col_a which are not in col_b by reversing the results of the isin() function.

Answer 5

df['Unique']=df.loc[~df['Ligand_miss'].isin(df['Ligand_hit']),'Ligand_miss'].reset_index(drop=True)

df
Out[624]: 
  Ligand_hit Ligand_miss  Unique
0     M00001      M00005  M00005
1     M00002      M00001  M00007
2     M00003      M00007     NaN
3     M00004      M00003     NaN