In [52]:

    

%pylab inline
from myfisher import fisherTestVec
import statsmodels.api as sm


    

    




Populating the interactive namespace from numpy and matplotlib

In [64]:

    

df = pd.read_csv(DATA_PATH + 'adevent.csv')
eCols = ['E%d' % i for i in range(1,29)]
df.loc[:,eCols] = df.loc[:,eCols].applymap(lambda s: 1. if s=='event' else 0.).astype(float)
N = df.shape[0]
print N


    

    




160

In [65]:

    

aInd = df.group == 'A'
bInd = df.group == 'B'

a = array([df[col].loc[bInd].sum() for col in eCols])
b = array([df[col].loc[aInd].sum() for col in eCols])
c = N - a
d = N - b

OR, p = fisherTestVec(a, b, c, d, alternative='greater')
resDf = pd.DataFrame(dict(OR=OR,p=p),index=eCols)


    

    
Out[65]:





80

In [62]:

    

h, resDf['bonferroni'], _, _ = sm.stats.multipletests(resDf['p'], method='bonferroni')
h, resDf['holm'], _, _ = sm.stats.multipletests(resDf['p'], method='holm')
h, resDf['fdr_bh'], _, _ = sm.stats.multipletests(resDf['p'], method='fdr_bh')


    

In [63]:

    

resDf = resDf.sort('p')
resDf.head()


    

    
Out[63]:






  

    

      

      
OR
      
p
      
bonferroni
      
holm
      
fdr_bh
    
  
  

    

      
E1
      
3.518519
      
0.001390
      
0.038927
      
0.038927
      
0.038927
    
    

      
E8
      
inf
      
0.030275
      
0.847700
      
0.817425
      
0.423850
    
    

      
E6
      
inf
      
0.061326
      
1.000000
      
1.000000
      
0.572379
    
    

      
E5
      
2.548387
      
0.223967
      
1.000000
      
1.000000
      
1.000000
    
    

      
E15
      
inf
      
0.249216
      
1.000000
      
1.000000
      
1.000000