#!/usr/bin/env python
# coding: utf-8

# ## Load ontology
# 
# First load an ontology
# 

# ## Loading a local ontology
# 
# Assume we have an ontology on disk in obographs-json format, e.g. nucleus.json.
# 
# First create an ontology *factory*, and create an ontology by passing the filename as the handle

# In[1]:


from ontobio.ontol_factory import OntologyFactory

ofactory = OntologyFactory()
ont = ofactory.create("nucleus.json") ## interpreted as a json local handle


# Now we can search the ontology by label;
# regexes can be used, but here we specify an exact match, so we expect one result:
#     

# In[2]:


[nucleus] = ont.search('nucleus')


# ### Ancestor Queries

# In[3]:


ont.ancestors(nucleus)


# In[4]:


## Extract labels for results
["{} '{}'".format(x, ont.label(x)) for x in ont.ancestors(nucleus)]


# In[5]:


## Filter ancestors based on relationship type
ont.ancestors(nucleus, relations='subClassOf')


# ### Node objects

# In[42]:


## Fetch the nucleus node

ont.node(nucleus)


# Note that the ontol object provides special convenience methods for accessing
# information in a node (see Synonyms section below). However, it is always possible to access
# the dictionary object for a node and to probe this directly, for  example:
# 

# In[44]:


ont.node(nucleus)['meta']['synonyms']


# ### Synonyms Objects
# 
# A synonym object includes the scope, type and any provenance for the synonym.
# See the notebook on lexical mapping for more details on ontology lexical elements

# In[34]:


ont.synonyms('GO:0005575')


# ### Networkx Graphs
# 
# The underlying representation of the ontology is a networkx graph. You can access a graph directly:

# In[37]:


graph = ont.get_graph()


# In[38]:


## number of nodes in this test ontology
len(graph)


# In[41]:


graph.node[nucleus]


# In[45]:


import networkx
import matplotlib.pyplot as plt


# In[47]:


networkx.draw(graph)


# In[ ]:


plt.savefig('foo.png')