In [20]:

    

import pandas as pd


    

In [23]:

    

from bokeh.io import push_notebook, show, output_notebook
from bokeh.plotting import figure
output_notebook()


    

    






    

        
        Loading BokehJS ...
    






    

In [24]:

    

df_rep82 = pd.read_csv("./../data/references/rep82/utree/rep82.gg.log", sep="\t", header=None, index_col=0)
df_rep82_dusted = pd.read_csv("./../data/references/rep82_dusted/utree/rep82_dusted.gg.log", sep="\t", header=None, index_col=0)


    

In [25]:

    

for df in (df_rep82, df_rep82_dusted):
    df["level"] = [_.count(";") + 1 for _ in df.index]


    

In [34]:

    

from bokeh.charts import Histogram
hist = Histogram(list(df_rep82['level']), title="Taxonomic Levels", bins=3)
hist2 = Histogram(list(df_rep82_dusted['level']), title="Taxonomic Levels", bins=3)


    

In [32]:

    

show(hist)


    

In [35]:

    

show(hist2)


    

In [40]:

    

# Only two levels? Is the log being output correctly?
print(df_rep82["level"].unique())
print(df_rep82_dusted["level"].unique())


    

    




[8 1]
[8 1]

In [39]:

    

levels.unique()


    

    
Out[39]:





array([8, 1])