KeyError when extracting data from a pandas.core.series.Series

Question

In the following ipython3 session, I read differently-formatted tables and make the sum of the values found in one of the columns:

In [278]: F = pd.read_table("../RNA_Seq_analyses/mapping_worm_number_tests/hisat2/mapped_C_elegans/feature_count/W100_1_on_C_elegans/protein_coding_fwd_counts.txt", skip
     ...: rows=2, usecols=[6]).sum()

In [279]: S = pd.read_table("../RNA_Seq_analyses/mapping_worm_number_tests/hisat2/mapped_C_elegans/intersect_count/W100_1_on_C_elegans/protein_coding_fwd_counts.txt", us
     ...: ecols=[6], header=None).sum()

In [280]: S
Out[280]: 
6    3551266
dtype: int64

In [281]: F
Out[281]: 
72    3164181
dtype: int64

In [282]: type(F)
Out[282]: pandas.core.series.Series

In [283]: type(S)
Out[283]: pandas.core.series.Series

In [284]: F[0]
Out[284]: 3164181

In [285]: S[0]
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-285-5a4339994a41> in <module>()
----> 1 S[0]

/home/bli/.local/lib/python3.6/site-packages/pandas/core/series.py in __getitem__(self, key)
    601             result = self.index.get_value(self, key)
    602 
--> 603             if not is_scalar(result):
    604                 if is_list_like(result) and not isinstance(result, Series):
    605 

/home/bli/.local/lib/python3.6/site-packages/pandas/indexes/base.py in get_value(self, series, key)

pandas/index.pyx in pandas.index.IndexEngine.get_value (pandas/index.c:3323)()

pandas/index.pyx in pandas.index.IndexEngine.get_value (pandas/index.c:3026)()

pandas/index.pyx in pandas.index.IndexEngine.get_loc (pandas/index.c:4009)()

pandas/src/hashtable_class_helper.pxi in pandas.hashtable.Int64HashTable.get_item (pandas/hashtable.c:8146)()

pandas/src/hashtable_class_helper.pxi in pandas.hashtable.Int64HashTable.get_item (pandas/hashtable.c:8090)()

KeyError: 0

How come the F and S objects have different behaviours if they result from similar operation (sum) and are of the same type (pandas.core.series.Series)?

What is the correct way to extract the value I want (the sum of a column)?

Edit: Trying solutions:

In [297]: F["72"]
Out[297]: 3164181

In [298]: S["6"]
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
pandas/index.pyx in pandas.index.IndexEngine.get_loc (pandas/index.c:4009)()

pandas/src/hashtable_class_helper.pxi in pandas.hashtable.Int64HashTable.get_item (pandas/hashtable.c:8125)()

TypeError: an integer is required

During handling of the above exception, another exception occurred:

KeyError                                  Traceback (most recent call last)
<ipython-input-298-0127424036a0> in <module>()
----> 1 S["6"]

/home/bli/.local/lib/python3.6/site-packages/pandas/core/series.py in __getitem__(self, key)
    601             result = self.index.get_value(self, key)
    602 
--> 603             if not is_scalar(result):
    604                 if is_list_like(result) and not isinstance(result, Series):
    605 

/home/bli/.local/lib/python3.6/site-packages/pandas/indexes/base.py in get_value(self, series, key)

pandas/index.pyx in pandas.index.IndexEngine.get_value (pandas/index.c:3323)()

pandas/index.pyx in pandas.index.IndexEngine.get_value (pandas/index.c:3026)()

pandas/index.pyx in pandas.index.IndexEngine.get_loc (pandas/index.c:4075)()

KeyError: '6'

Further investigating:

In [306]: print(S.index)
Int64Index([6], dtype='int64')

In [307]: print(F.index)
Index(['72'], dtype='object')

In [308]: S[6]
Out[308]: 3551266

So the two objects ended up having different types of indices. This kind of behaviour reminds me of R...

It seems that header=None resulted in columns indexed by numbers for S, whereas the absence of header=None combined with skiprows=2 resulted in the index being generated from data read on the third row. (And this revealed a bug in the way I parsed the data in pandas...)

Answer 1

I think you need:

#select first value of one element series
f = F.iat[0]
#alternative 
#f = F.iloc[0]

Or:

#convert to numpy array and select first value
f = F.values[0]

Or:

f = F.item()

And I think you get error, because no index value 0.

As IanS commented should be working select by index value 6 and 72:

f = F[72] 
#f = f.loc[72]

s = S[6]
#s = S.loc[6]

Sample:

F = pd.Series([3164181], index=[72])

f = F[72] 
print (f)
3164181

print (F.index)
Int64Index([72], dtype='int64')

print (F.index.tolist())
[72]

f = F[0] 
print (f)

KeyError: 0

You get one integer index in S, because parameter header=None - pandas add default index (0,1,...). For F is used 6th column called '72' - it is string. There is difference.