from ontobio.ontol_factory import OntologyFactory

# Create ontology object, for GO
# Transparently uses remote SPARQL service.
# (May take a few seconds to run first time, Jupyter will show '*'. BE PATIENT, do
# not re-execute cell)
ofactory = OntologyFactory()
ont = ofactory.create('go')

from ontobio.assoc_factory import AssociationSetFactory

MOUSE = 'NCBITaxon:10090'

# Create association set
# Transparently uses remote Monarch service.
# (May take a few seconds to run first time, Jupyter will show '*'. BE PATIENT, do
# not re-execute cell)
afactory = AssociationSetFactory()
aset = afactory.create(ontology=ont,
                       subject_category='gene',
                       object_category='function',
                       taxon=MOUSE)

aset.subjects[0:5]

aset.jaccard_similarity('MGI:1929214', 'MGI:1917826')

HUMAN = 'NCBITaxon:9606'

# Create association set
# Transparently uses remote Monarch service.
# (May take a few seconds to run first time, Jupyter will show '*'. BE PATIENT, do
# not re-execute cell)
afactory = AssociationSetFactory()
aset_human = afactory.create(ontology=ont,
                       subject_category='gene',
                       object_category='function',
                       taxon=HUMAN)

ZFISH = 'NCBITaxon:7955'

# Create association set
# Transparently uses remote Monarch service.
# (May take a few seconds to run first time, Jupyter will show '*'. BE PATIENT, do
# not re-execute cell)
afactory = AssociationSetFactory()
aset_zfish = afactory.create(ontology=ont,
                       subject_category='gene',
                       object_category='function',
                       taxon=ZFISH)

# TODO - do not re-read
aset = afa.create_from_gaf(url, ontology=ont)



from collections import defaultdict
rmap = defaultdict(set)
for a in assocs:
    ev = a['evidence']['type']
    gene = a['subject']['id']
    p = a['object']['id']
    up = process_regulates(p)
    for rp in up:
        rmap[(gene,rp)].add('R')
    rmap[(gene,p)].add('I')
    

len(rmap)

mixed = [(k,v) for (k,v) in rmap.items() if len(v)>1]
len(mixed)

mixed[0:20]

ont.label('GO:0007049')