I have a data frame like below:
s1 AA AG AG GG AA
s2 DI DD GG GG GG
S3 TT CC TC TT TC
S3 AA DI AA AA AA
S3 CC CC DD CC CC
and I want to replace all other strings to II in the row, if there is DI or DD in the row. so final dataframe should be like this.
s1 AA AG AG GG AA
s2 DI DD II II II
S3 TT CC TC TT TC
S3 II DI II II II
S4 II II DD II II
Any suggestions is appreciated. Thank you in advance
Check both values by DataFrame.isin and then replace only rows with at least one True with DataFrame.where:
m = df.isin(['DI','DD'])
m1 = m.any(1)
df[m1] = df[m1].where(m, 'II')
Or use numpy broadcasting for chain both masks:
m = df.isin(['DI','DD'])
df = df.where(m.values | ~m.any(1).values[:, None], 'II')
print (df)
0 1 2 3 4 5
0 s1 AA AG AG GG AA
1 II DI DD II II II
2 S3 TT CC TC TT TC
3 II II DI II II II
4 II II II DD II II
EDIT:
Working with - is possible by chained mask by | for bitwise OR:
print (df)
0 1 2 3 4 5
0 s1 AA AG AG GG -
1 s2 DI DD GG GG -
2 S3 TT CC TC TT TC
3 S3 AA DI AA AA AA
4 S3 CC CC DD CC CC
m = df.isin(['DI','DD'])
m1 = m.any(1)
df[m1] = df[m1].where(m | df.eq('-'), 'II')
print (df)
0 1 2 3 4 5
0 s1 AA AG AG GG -
1 II DI DD II II -
2 S3 TT CC TC TT TC
3 II II DI II II II
4 II II II DD II II
m = df.isin(['DI','DD', '-']) ?m = df.replace('-', np.nan).isin(['DI','DD']) ?m = df.isin(['DI','DD']).sum(axis=1).astype(bool)
df.loc[m,:] = df.loc[m,:].applymap(lambda x: x if x in ["DI","DD"] else "II")
print(df)
