def bioconcepts2pubtator_annotations(tag, index):
"""Bioconcepts to Annotations
Specifically for bioconcepts2pubtator and converts each annotation
into an annotation object that BioC can parse.
Keyword Arguments:
tag -- the annotation line that was parsed into an array
index -- the id of each document specific annotation
"""
annt = BioCAnnotation()
annt.id = str(index)
annt.infons["type"] = tag["type"]
# If the annotation type is a Gene,Species, Mutation, SNP
# Write out relevant tag
tag_type = tag['type'] or ''
tag_id = tag['tag_id']
if tag_type == "Gene":
annt.infons["NCBI Gene"] = tag_id
elif tag_type == "Species":
annt.infons["NCBI Species"] = tag_id
elif "Mutation" in tag_type:
annt.infons["tmVar"] = tag_id
elif "SNP" in tag_type:
annt.infons["tmVar"] = tag_id
else:
# If there is no MESH ID for an annotation
if tag_id:
# check to see if there are multiple mesh tags
if "|" in tag_id:
# Write out each MESH id as own tag
for tag_num, ids in enumerate(tag_id.split("|")):
# Some ids dont have the MESH:#### form so added case to that
if ":" not in ids:
annt.infons["MESH {}".format(tag_num)] = tag_id
else:
term_type, term_id = ids.split(":")
annt.infons["{} {}".format(term_type,
tag_num)] = term_id
else:
# Some ids dont have the MESH:#### form so added case to that
if ":" in tag_id:
term_type, term_id = tag_id.split(":")
annt.infons[term_type] = term_id
else:
annt.infons["MESH"] = tag_id
else:
annt.infons["MESH"] = "Unknown"
location = BioCLocation()
location.offset = str(tag["start"])
location.length = str(len(tag["term"]))
annt.locations.append(location)
annt.text = tag["term"]
return annt
def add_annotation(triple, annotation_id):
# initialize annotation element
bioc_annotation = BioCAnnotation()
# MeSH term in a tag <text> ... </text> (origininal term, searched case insensitive)
bioc_annotation.text = triple[2]
# generate XML structure for the annotation and add infon
bioc_annotation.id = str(annotation_id)
bioc_annotation.put_infon('type', 'MeSH term')
# add location element
bioc_location = BioCLocation()
# add length of MeSH term
bioc_location.length = str(triple[1])
# add start position (offset)
bioc_location.offset = str(triple[0])
bioc_annotation.add_location(bioc_location)
return bioc_annotation
def add_annotation(triple, annotation_id):
# initialize annotation element
bioc_annotation = BioCAnnotation()
# MeSH term in a tag <text> ... </text> (origininal term, searched case insensitive)
bioc_annotation.text = triple[2]
# generate XML structure for the annotation and add infon
bioc_annotation.id = str(annotation_id)
bioc_annotation.put_infon('type', 'MeSH term')
# add location element
bioc_location = BioCLocation()
# add length of MeSH term
bioc_location.length = str(triple[1])
# add start position (offset)
bioc_location.offset = str(triple[0])
bioc_annotation.add_location(bioc_location)
return bioc_annotation
def brat2bioc_entity(bratentity: BratEntity) -> BioCAnnotation:
ann = BioCAnnotation()
ann.id = bratentity.id
ann.text = bratentity.text
ann.infons['type'] = bratentity.type
for span in bratentity.locations:
ann.add_location(BioCLocation(span.begin, span.end - span.begin))
return ann
def to_bioc(self):
entity_bioc = BioCAnnotation()
entity_bioc.infons['type'] = self.type
entity_bioc.text = self.text
entity_bioc.id = str(self.id)
location = BioCLocation(self.start, len(self.text))
entity_bioc.add_location(location)
return entity_bioc
def __parse_annotation(self, tree):
annotation = BioCAnnotation()
annotation.id = tree.attrib['id']
annotation.infons = self.__parse_infons(tree)
annotation.text = tree.findtext('text')
for child in tree.findall('location'):
annotation.add_location(
BioCLocation(int(child.attrib['offset']), int(child.attrib['length'])))
return annotation
def __read_annotation(self, start_elem):
ann = BioCAnnotation()
ann.id = start_elem.get('id')
while self.__has_next():
event, elem = self.__next_event()
if event == 'start':
pass
elif event == 'end':
if elem.tag == 'text':
ann.text = elem.text
elif elem.tag == 'infon':
ann.infons[elem.get('key')] = elem.text
elif elem.tag == 'location':
ann.add_location(BioCLocation(int(elem.get('offset')), int(elem.get('length'))))
elif elem.tag == 'annotation':
return ann
raise RuntimeError("should not reach here") # pragma: no cover