Count the number of leaf nodes for each HPO category under Phenotypic Abnormality (HP:0000118)

count-leafs.py

from neo4j import GraphDatabase
from jinja2 import Template
import json

QUERY="""
match p=(d:DATA)-[:PAYLOAD]->(n:S_HP)-[e:R_subClassOf*0..11]->(m:S_HP)-[e2:R_subClassOf]->(:S_HP)<-[:PAYLOAD]-(z) 
where z.notation = 'HP:0000118' 
and d.notation='{{ leaf }}' 
and all(x in e where x.source=n.source or n.source in x.source) 
and (e2.source=n.source or n.source in e2.source) 
with d,m match (m)<-[:PAYLOAD]-(z) 
return distinct z.notation as ID, z.label as LABEL
"""

uri = "bolt://disease.ncats.io:80"
driver = GraphDatabase.driver(uri, auth=("neo4j", ""))
template = Template(QUERY)

def run_cypher(tx, query):
    leafs = []
    for row in tx.run(query):
        leafs.append((row['ID'], row['LABEL']))
    return leafs

def get_leafs():
    CYPHER="""
match p=(d)-[:PAYLOAD]->(n:S_HP) 
where not (n)<-[:R_subClassOf]-() 
and not n:TRANSIENT 
and not n:AnnotationProperty 
and not n:ObjectProperty 
and d.notation =~ 'HP:.*'
return d.notation as ID, d.label as LABEL
order by d.notation desc
"""
    with driver.session() as session:
        return session.read_transaction(run_cypher, CYPHER)

def get_categories(leaf):
    with driver.session() as session:
        return session.read_transaction(run_cypher, template.render(leaf=leaf))

if __name__ == '__main__':
    leafs = get_leafs()
    categories = {}
    for l in leafs:
        cats = get_categories(l[0])
        #print ('%s: %s' % l)
        for cat in cats:
            #print ('... %s %s' % cat)
            if cat[0] not in categories:
                categories[cat[0]] = {
                    'label': cat[1],
                    'count': 1
                    }
            else:
                c = categories[cat[0]]['count']
                categories[cat[0]]['count'] = c+1
    print (json.dumps(categories, indent=2))

This script can be replaced with this cypher query!

match (d)-[:PAYLOAD]->(n:S_HP) 
where not (n)<-[:R_subClassOf]-() 
and not n:TRANSIENT 
and not n:AnnotationProperty 
and not n:ObjectProperty 
and d.notation =~ 'HP:.*' with d as leaf 
match (d:DATA)-[:PAYLOAD]->(n:S_HP)-[e:R_subClassOf*0..11]->(m:S_HP)-[e2:R_subClassOf]->(:S_HP)<-[:PAYLOAD]-(z) 
where z.notation = 'HP:0000118' 
and d.notation= leaf.notation
and all(x in e where x.source=n.source or n.source in x.source) 
and (e2.source=n.source or n.source in e2.source) 
with leaf,m match (m)<-[:PAYLOAD]-(z) 
return z.notation as ID, z.label as LABEL,count(distinct leaf) as COUNT order by COUNT desc