In [7]:

    

import networkx as nx
import custom_funcs as cf
import pandas as pd


    

In [5]:

    

# Read in the graph data and clean it.
G = nx.read_gpickle('20150902_all_ird Final Graph.pkl')
G = cf.clean_host_species_names(G)
G = cf.impute_reassortant_status(G)
G = cf.impute_weights(G)
G = cf.remove_zero_weighted_edges(G)


    

In [11]:

    

# Read in host groups data
df = pd.read_csv('supp_data/host_groups.csv')

# How many wild bird species labels are there?
len(df[(df['Habitat/setting'] == 'Wild')]['Species'].drop_duplicates())


    

    
Out[11]:





142

In [12]:

    

142 ** 2


    

    
Out[12]:





20164

In [13]:

    

len(G.nodes())


    

    
Out[13]:





18632

In [14]:

    

len(G.edges())


    

    
Out[14]:





25683

In [ ]: