熊猫csv / multiindex子集

Question

I am trying to read a .csv file that has two rows of header information as a multiindex, so that I can later access a column given 2 identifiers. The file looks like this (tab-delimited), and the values that are NA are deliberately that way:

ind Human Human Human Mouse Mouse Mouse ...
(null) Codon Freq minmax Codon Freq minmax ...
0 ATG 12.5 -5.2 --- NA NA ...
1 AAA 8.9 -25.5 --- NA NA ...
2 GGA 16.5 12.4 ATG 11.9 6.5 ...

I can read the file in with two rows of headers, but this results in an object of class 'pandas.core.frame.DataFrame' instead of 'pandas.core.index.MultiIndex':
data = pd.read_csv('alignment.csv', sep="\\t", header=[0,1])

When I try specifying index_col=0, as some examples in the documentation do, I get a "IndexError: list index out of range" error, which was a solution to several related questions but for some reason isn't working for me.

Moving on, I've attempted to subset the data in a variety of ways, all of which have failed. The closest I've gotten (I think) to what I want is by doing
temp = data.ix[:,[("","ind"),("Human","minmax")]]
...which at least gives me a DataFrame of the right dimensions and labeled correctly, but all of the values have been replaced with NaN. Using .loc gives me an error about being improperly sorted, and I haven't been able to get .xs to work at all.

Essentially I'm looking for a way to subset the data set based on the species and the parameter (eg human and minmax). I've looked through several related questions here but haven't been able to solve the problem yet. How could I achieve this?

Answer 1

Hmm... it seems to work for me... what version of Pandas/Python are you using?

df= pd.read_clipboard(header = [0,1], index_col=0)

df
Out[389]: 
ind    Human              Mouse               ...
(null) Codon  Freq minmax Codon  Freq minmax  ...
0        ATG  12.5   -5.2   ---   NaN    NaN  ...
1        AAA   8.9  -25.5   ---   NaN    NaN  ...
2        GGA  16.5   12.4   ATG  11.9    6.5  ...


df.Human.minmax
df.Human.minmax
Out[390]: 
0    -5.2
1   -25.5
2    12.4
Name: minmax, dtype: float64