only with pd.eval() you can solve your problem:
df.eval('f_1=1_2+1_3+1_4',inplace=True)
Do it for every column you want to add