def create_ontology_map(golr_base_url):
ontology = utils.golr_fetch(golr_base_url, select_ontology)
ontology = ontology['response']['docs']
map = {}
for item in ontology:
map[item['annotation_class']] = item
return map
def golr_fetch_bioentities_taxon(taxon):
url = "select?fq=document_category:%22bioentity%22&q=*:*&wt=json&rows=0&facet=true&facet.field=type&facet.field=taxon&facet.limit=1000000&facet.mincount=1&fq=taxon:\"" + taxon + "\""
response = utils.golr_fetch(golr_base_url, url)
# multiple queries: a bit complicated but necessary due to solr 3.6 unable to do composite faceting and for speed considerations
# * can indicate the is_a closure to find the stats on that specific aspect
# * if evidence code was present, we could use a similar strategy
url_bp = "select?fq=document_category:%22bioentity%22&q=*:*&wt=json&facet=true&facet.field=type&facet.field=taxon&facet.limit=1000000&facet.mincount=1&rows=0&fq=taxon:\"" + taxon + "\"&fq=isa_partof_closure:\"" + BP + "\""
response_bp = utils.golr_fetch(golr_base_url, url_bp)
url_mf = "select?fq=document_category:%22bioentity%22&q=*:*&wt=json&facet=true&facet.field=type&facet.field=taxon&facet.limit=1000000&facet.mincount=1&rows=0&fq=taxon:\"" + taxon + "\"&fq=isa_partof_closure:\"" + MF + "\""
response_mf = utils.golr_fetch(golr_base_url, url_mf)
url_cc = "select?fq=document_category:%22bioentity%22&q=*:*&wt=json&facet=true&facet.field=type&facet.field=taxon&facet.limit=1000000&facet.mincount=1&rows=0&fq=taxon:\"" + taxon + "\"&fq=isa_partof_closure:\"" + CC + "\""
response_cc = utils.golr_fetch(golr_base_url, url_cc)
return {ALL: response, BP: response_bp, MF: response_mf, CC: response_cc}
def golr_fetch_annotation_by_evidence_by_species(taxon, exclude_pb_only):
options = ""
if exclude_pb_only:
options = "&fq=!annotation_class:\"GO:0005515\""
url = 'select?fq=document_category:%22annotation%22&q=*:*&wt=json&fq=taxon:%22' + taxon + '%22&facet=true&facet.field=evidence_type&facet.limit=10000&rows=0' + options
response = utils.golr_fetch(golr_base_url, url)
url_bp = 'select?fq=document_category:%22annotation%22&q=*:*&wt=json&fq=taxon:%22' + taxon + '%22&facet=true&facet.field=evidence_type&facet.limit=10000&rows=0&fq=isa_partof_closure:\"' + BP + '\"' + options
response_bp = utils.golr_fetch(golr_base_url, url_bp)
url_mf = 'select?fq=document_category:%22annotation%22&q=*:*&wt=json&fq=taxon:%22' + taxon + '%22&facet=true&facet.field=evidence_type&facet.limit=10000&rows=0&fq=isa_partof_closure:\"' + MF + '\"' + options
response_mf = utils.golr_fetch(golr_base_url, url_mf)
url_cc = 'select?fq=document_category:%22annotation%22&q=*:*&wt=json&fq=taxon:%22' + taxon + '%22&facet=true&facet.field=evidence_type&facet.limit=10000&rows=0&fq=isa_partof_closure:\"' + CC + '\"' + options
response_cc = utils.golr_fetch(golr_base_url, url_cc)
return {ALL: response, BP: response_bp, MF: response_mf, CC: response_cc}
def get_references():
refs = utils.golr_fetch(golr_base_url, golr_select_references)
refs = utils.build_map(refs['facet_counts']['facet_fields']['reference'])
return refs
def compute_stats(golr_url, release_date, exclude_pb_only=False):
"""
compute stats on GO annotations - can specify if we include or exclude annotations to protein binding only
"""
global golr_base_url
golr_base_url = golr_url
print("Will use golr url: ", golr_base_url)
print("1 / 4 - Fetching GO terms...")
all_terms = utils.golr_fetch(golr_base_url, golr_select_ontology)
print("Done.")
print("2 / 4 - Fetching GO annotations...")
if exclude_pb_only:
all_annotations = utils.golr_fetch(
golr_base_url, golr_select_annotations_no_pbinding)
else:
all_annotations = utils.golr_fetch(golr_base_url,
golr_select_annotations)
print("Done.")
print("3 / 4 - Fetching GO bioentities...")
all_entities = utils.golr_fetch(golr_base_url, golr_select_bioentities)
# we have to manually update the facts of the first query if we want to remove the bioentities annotated only to protein binding
if exclude_pb_only:
all_entities_no_pb = utils.golr_fetch(golr_base_url,
golr_select_bioentities_pb)
# print(all_entities_no_pb)
entities_type_no_pb = {}
entities_taxon_no_pb = {}
count = 0
for doc in all_entities_no_pb['response']['docs']:
if len(doc['annotation_class_list']) > 1:
continue
count += 1
if doc['type'] in entities_type_no_pb:
entities_type_no_pb[doc['type']] += 1
else:
entities_type_no_pb[doc['type']] = 1
if doc['taxon'] in entities_type_no_pb:
entities_taxon_no_pb[doc['taxon']] += 1
else:
entities_taxon_no_pb[doc['taxon']] = 1
# finally update the type facet field
types = all_entities['facet_counts']['facet_fields']['type']
for i in range(0, len(types), 2):
ctype = types[i]
retr_value = entities_type_no_pb[
ctype] if ctype in entities_type_no_pb else 0
types[i + 1] = types[i + 1] - retr_value
all_entities['facet_counts']['facet_fields']['type'] = types
all_entities['response'][
'numFound'] = all_entities['response']['numFound'] - count
# and update the taxon facet field
taxons = all_entities['facet_counts']['facet_fields']['taxon']
for i in range(0, len(taxons), 2):
ctaxon = taxons[i]
retr_value = entities_taxon_no_pb[
ctaxon] if ctaxon in entities_taxon_no_pb else 0
taxons[i + 1] = taxons[i + 1] - retr_value
all_entities['facet_counts']['facet_fields']['taxon'] = taxons
print("Done.")
qualifiers = utils.golr_fetch(golr_base_url, golr_select_qualifiers)
qualifiers = utils.build_map(
qualifiers['facet_counts']['facet_fields']['qualifier'])
print("4 / 4 - Creating Stats...")
prepare_globals(all_annotations)
print("\t4a - globals prepared")
stats = create_stats(all_terms, all_annotations, all_entities,
release_date, qualifiers, exclude_pb_only)
print("Done.")
return stats
def golr_fetch_references_group(group):
url = "select?fq=document_category:%22annotation%22&q=*:*&wt=json&rows=0&facet.limit=10000000&facet.mincount=1&facet=true&facet.field=reference&fq=assigned_by:\"" + group + "\""
response = utils.golr_fetch(golr_base_url, url)
return response