Come ordinare e mantenere gli indici comuni da due DataFrames

Question

ho due DataFrames:

import pandas as pd 
import io 
from scipy import stats 


ctrl=u"""probegenes,sample1,sample2,sample3 
1415777_at Pnliprp1,20,0.00,11 
1415884_at Cela3b,47,0.00,100 
1415805_at Clps,17,0.00,55 
1115805_at Ckkk,77,10.00,5.5 
""" 

df_ctrl = pd.read_csv(io.StringIO(ctrl),index_col='probegenes') 

test=u"""probegenes,sample1,sample2,sample3 
1415777_at Pnliprp1,20.1,10.00,22.3 
1415805_at Clps,7,3.00,1.5 
1415884_at Cela3b,47,2.01,30""" 

df_test = pd.read_csv(io.StringIO(test),index_col='probegenes') 

Sembrano questo:

In [35]: df_ctrl 
Out[35]: 
        sample1 sample2 sample3 
probegenes 
1415777_at Pnliprp1  20  0  11.0 
1415884_at Cela3b   47  0 100.0 
1415805_at Clps   17  0  55.0 
1115805_at Ckkk   77  10  5.5 

In [36]: df_test 
Out[36]: 
        sample1 sample2 sample3 
probegenes 
1415777_at Pnliprp1  20.1 10.00  22.3 
1415805_at Clps   7.0  3.00  1.5 
1415884_at Cela3b  47.0  2.01  30.0 

mi piacerebbe:

1. Get un comune index per entrambi DataFrame
2. Reo sono entrambi identici allo DataFrame.

Quindi, alla fine ottengo due nuovi DataFrame:

new_df_ctrl 

        sample1 sample2 sample3 
probegenes 
1415884_at Cela3b   47  0 100.0 
1415805_at Clps   17  0  55.0 
1415777_at Pnliprp1  20  0  11.0 


new_df_test 

        sample1 sample2 sample3 
probegenes 
1415884_at Cela3b  47.0  2.01  30.0 
1415805_at Clps   7.0  3.00  1.5 
1415777_at Pnliprp1  20.1 10.00  22.3 

Answer 1

È possibile utilizzare join con il parametro how='inner' per ottenere l'indice comune. Quindi basta reindicizzare ogni dataframe utilizzando questo indice comune.

idx = df_ctrl.join(df_test, rsuffix='_', how='inner').index 

>>> df_ctrl.reindex(idx) 
        sample1 sample2 sample3 
probegenes          
1415777_at Pnliprp1  20  0  11 
1415805_at Clps   17  0  55 
1415884_at Cela3b   47  0  100 

>>> df_test.reindex(idx) 
        sample1 sample2 sample3 
probegenes          
1415777_at Pnliprp1  20.1 10.00  22.3 
1415805_at Clps   7.0  3.00  1.5 
1415884_at Cela3b  47.0  2.01  30.0 

Answer 2

Si potrebbe utilizzare pd.Index.intersection() e selezionare con .loc[] o .reindex(). Utilizzare .sort_values() per il numero index per ottenere l'ordine desiderato:

idx = df_ctrl.index.intersection(df_test.index).sort_values(ascending=False) 

df_ctrl.loc[idx] 

        sample1 sample2 sample3 
probegenes          
1415884_at Cela3b   47  0.0 100.0 
1415805_at Clps   17  0.0  55.0 
1415777_at Pnliprp1  20  0.0  11.0 

df_test.loc[idx] 

        sample1 sample2 sample3 
probegenes          
1415884_at Cela3b  47.0  2.01  30.0 
1415805_at Clps   7.0  3.00  1.5 
1415777_at Pnliprp1  20.1 10.00  22.3