def add_association_to_graph(self):
"""
The reified relationship between a disease and a phenotype is decorated
with some provenance information.
This makes the assumption that both the disease and phenotype
are classes.
:param g:
:return:
"""
# add the basic association nodes
# if rel == self.globaltt[['has disposition']:
Assoc.add_association_to_graph(self)
# anticipating trouble with onsets ranges that look like curies
if self.onset is not None and self.onset != '':
self.graph.addTriple(self.assoc_id, self.globaltt['onset'], self.onset)
if self.frequency is not None and self.frequency != '':
self.graph.addTriple(
self.assoc_id, self.globaltt['frequency'], self.frequency)
return
def add_association_to_graph(self, association_category=None):
"""
The reified relationship between a disease and a phenotype is decorated
with some provenance information.
This makes the assumption that both the disease and phenotype
are classes.
:param g:
:param disease_category: a biolink category CURIE for disease_id (defaults to
biolink:Disease via the constructor)
:param phenotype_category: a biolink category CURIE for phenotype_id (defaults
to biolink:PhenotypicFeature via the constructor)
:return:
"""
# add the basic association nodes
# if rel == self.globaltt[['has disposition']:
Assoc.add_association_to_graph(self)
# anticipating trouble with onsets ranges that look like curies
if self.onset is not None and self.onset != '':
self.graph.addTriple(self.assoc_id, self.globaltt['onset'], self.onset,
object_category=self.object_category)
if self.frequency is not None and self.frequency != '':
self.graph.addTriple(
self.assoc_id, self.globaltt['frequency'], self.frequency,
object_category=self.object_category)
return
def make_association(self, record):
"""
contstruct the association
:param record:
:return: modeled association of genotype to mammalian phenotype
"""
model = Model(self.graph)
record['relation']['id'] = self.resolve("has phenotype")
# define the triple
gene = record['subject']['id']
relation = record['relation']['id']
phenotype = record['object']['id']
# instantiate the association
g2p_assoc = Assoc(self.graph,
self.name,
sub=gene,
obj=phenotype,
pred=relation)
# add the references
references = record['evidence']['has_supporting_reference']
# created RGDRef prefix in curie map to route to proper reference URL in RGD
references = [
x.replace('RGD', 'RGDRef') if 'PMID' not in x else x
for x in references
]
if len(references) > 0:
# make first ref in list the source
g2p_assoc.add_source(identifier=references[0])
ref_model = Reference(self.graph, references[0],
self.globaltt['publication'])
ref_model.addRefToGraph()
if len(references) > 1:
# create equivalent source for any other refs in list
# This seems to be specific to this source and
# there could be non-equivalent references in this list
for ref in references[1:]:
model.addSameIndividual(sub=references[0], obj=ref)
# add the date created on
g2p_assoc.add_date(date=record['date'])
g2p_assoc.add_evidence(self.resolve(
record['evidence']['type'])) # ?set where?
g2p_assoc.add_association_to_graph()
return
def add_association_to_graph(self,
entity_category=None,
phenotype_category=None):
"""
Overrides Association by including bnode support
The reified relationship between a genotype (or any genotype part)
and a phenotype is decorated with some provenance information.
This makes the assumption that
both the genotype and phenotype are classes.
currently hardcoded to map the annotation to the monarch namespace
:param g:
:param entity_category: a biolink category CURIE for self.sub
:param phenotype_category: a biolink category CURIE for self.obj
:return:
"""
# is this kosher?
Assoc.add_association_to_graph(self)
# make a blank stage
if self.start_stage_id or self.end_stage_id is not None:
stage_process_str = '-'.join(
(str(self.start_stage_id), str(self.end_stage_id)))
stage_process_id = ':'.join( # bnode
('_', self.gut.digest_id(stage_process_str)))
self.model.addIndividualToGraph(
stage_process_id, None, self.globaltt['developmental_process'])
self.graph.addTriple(stage_process_id, self.globaltt['label'],
stage_process_str)
self.graph.addTriple(stage_process_id,
self.globaltt['starts during'],
self.start_stage_id)
self.graph.addTriple(stage_process_id,
self.globaltt['ends during'],
self.end_stage_id)
self.stage_process_id = stage_process_id
self.graph.addTriple(self.assoc_id, self.globaltt['has_qualifier'],
self.stage_process_id)
if self.environment_id is not None:
self.graph.addTriple(self.assoc_id, self.globaltt['has_qualifier'],
self.environment_id)
def _add_component_pathway_association(
self, eco_map, component, component_prefix, pathway_id,
pathway_prefix, pathway_label, go_ecode):
pathway = Pathway(self.graph)
pathway_curie = "{}:{}".format(pathway_prefix, pathway_id)
gene_curie = "{}:{}".format(component_prefix, component.strip())
eco_curie = eco_map[go_ecode]
pathway.addPathway(pathway_curie, pathway_label)
pathway.addComponentToPathway(gene_curie, pathway_curie)
association = Assoc(self.graph, self.name)
association.sub = gene_curie
association.rel = pathway.object_properties['involved_in']
association.obj = pathway_curie
association.set_association_id()
association.add_evidence(eco_curie)
association.add_association_to_graph()
return
def _parse_aeolus_data(self, document, or_limit=None):
model = Model(self.graph)
rxcui_curie = "RXCUI:{}".format(document['aeolus']['rxcui'])
uni_curie = "UNII:{}".format(document['aeolus']['unii'])
model.addLabel(rxcui_curie, document['aeolus']['drug_name'])
model.addLabel(uni_curie, document['aeolus']['drug_name'])
model.addSameIndividual(rxcui_curie, uni_curie)
self.graph.addTriple(rxcui_curie,
model.annotation_properties['inchi_key'],
document['unii']['inchikey'],
object_is_literal=True)
if or_limit is not None:
outcomes = (outcome for outcome in document['aeolus']['outcomes']
if 'ror' in outcome and outcome['ror'] >= or_limit)
else:
outcomes = (outcome for outcome in document['aeolus']['outcomes'])
for outcome in outcomes:
drug2outcome_assoc = Assoc(self.graph, self.name)
meddra_curie = "MEDDRA:{}".format(outcome['code'])
model.addLabel(meddra_curie, outcome['name'])
drug2outcome_assoc.sub = rxcui_curie
drug2outcome_assoc.obj = meddra_curie
drug2outcome_assoc.rel = Assoc.object_properties[
'causes_or_contributes']
drug2outcome_assoc.description = \
"A proportional reporting ratio or odds " \
"ratio greater than or equal to {} in the " \
"AEOLUS data was the significance cut-off " \
"used for creating drug-outcome associations".format(or_limit)
drug2outcome_assoc.add_association_to_graph()
drug2outcome_assoc.add_predicate_object(
Assoc.annotation_properties['probabalistic_quantifier'],
outcome['ror'], 'Literal')
self._add_outcome_evidence(drug2outcome_assoc.assoc_id, outcome)
self._add_outcome_provenance(drug2outcome_assoc.assoc_id, outcome)
def make_association(self, record):
"""
contstruct the association
:param record:
:return: modeled association of genotype to mammalian phenotype
"""
model = Model(self.graph)
record['relation']['id'] = self.resolve("has phenotype")
# define the triple
gene = record['subject']['id']
relation = record['relation']['id']
phenotype = record['object']['id']
# instantiate the association
g2p_assoc = Assoc(self.graph, self.name, sub=gene, obj=phenotype, pred=relation)
# add the references
references = record['evidence']['has_supporting_reference']
# created RGDRef prefix in curie map to route to proper reference URL in RGD
references = [
x.replace('RGD', 'RGDRef') if 'PMID' not in x else x for x in references]
if len(references) > 0:
# make first ref in list the source
g2p_assoc.add_source(identifier=references[0])
ref_model = Reference(
self.graph, references[0],
self.globaltt['publication']
)
ref_model.addRefToGraph()
if len(references) > 1:
# create equivalent source for any other refs in list
# This seems to be specific to this source and
# there could be non-equivalent references in this list
for ref in references[1:]:
model.addSameIndividual(sub=references[0], obj=ref)
# add the date created on
g2p_assoc.add_date(date=record['date'])
g2p_assoc.add_evidence(self.resolve(record['evidence']['type'])) # ?set where?
g2p_assoc.add_association_to_graph()
return
def add_association_to_graph(self):
"""
Overrides Association by including bnode support
The reified relationship between a genotype (or any genotype part)
and a phenotype is decorated with some provenance information.
This makes the assumption that
both the genotype and phenotype are classes.
currently hardcoded to map the annotation to the monarch namespace
:param g:
:return:
"""
Assoc.add_association_to_graph(self)
# make a blank stage
if self.start_stage_id or self.end_stage_id is not None:
stage_process_id = '-'.join(
(str(self.start_stage_id), str(self.end_stage_id)))
stage_process_id = '_:' + re.sub(r':', '', stage_process_id)
self.model.addIndividualToGraph(
stage_process_id, None, self.globaltt['developmental_process'])
self.graph.addTriple(stage_process_id,
self.globaltt['starts_during'],
self.start_stage_id)
self.graph.addTriple(stage_process_id,
self.globaltt['ends_during'],
self.end_stage_id)
self.stage_process_id = stage_process_id
self.graph.addTriple(self.assoc_id, self.globaltt['has_qualifier'],
self.stage_process_id)
if self.environment_id is not None:
self.graph.addTriple(self.assoc_id, self.globaltt['has_qualifier'],
self.environment_id)
return
def add_association_to_graph(self):
"""
Overrides Association by including bnode support
The reified relationship between a genotype (or any genotype part)
and a phenotype is decorated with some provenance information.
This makes the assumption that
both the genotype and phenotype are classes.
currently hardcoded to map the annotation to the monarch namespace
:param g:
:return:
"""
Assoc.add_association_to_graph(self)
# make a blank stage
if self.start_stage_id or self.end_stage_id is not None:
stage_process_id = '-'.join((str(self.start_stage_id),
str(self.end_stage_id)))
stage_process_id = '_:'+re.sub(r':', '', stage_process_id)
self.model.addIndividualToGraph(
stage_process_id, None, self.globaltt['developmental_process'])
self.graph.addTriple(
stage_process_id, self.globaltt['starts during'], self.start_stage_id)
self.graph.addTriple(
stage_process_id, self.globaltt['ends during'], self.end_stage_id)
self.stage_process_id = stage_process_id
self.graph.addTriple(
self.assoc_id, self.globaltt['has_qualifier'], self.stage_process_id)
if self.environment_id is not None:
self.graph.addTriple(
self.assoc_id, self.globaltt['has_qualifier'], self.environment_id)
return
def _add_gene_anatomy_association(self, gene_id, anatomy_curie, rank):
"""
:param gene_id: str Non curified ID
:param gene_label: str Gene symbol
:param anatomy_curie: str curified anatomy term
:param rank: str rank
:return: None
"""
g2a_association = Assoc(self.graph, self.name)
model = Model(self.graph)
gene_curie = "ENSEMBL:{}".format(gene_id)
rank = re.sub(r',', '', str(rank)) # ? can't do RE on a float ...
model.addIndividualToGraph(gene_curie, None)
g2a_association.sub = gene_curie
g2a_association.obj = anatomy_curie
g2a_association.rel = self.globaltt['expressed in']
g2a_association.add_association_to_graph()
g2a_association.add_predicate_object(
self.globaltt['has_quantifier'], float(rank), 'Literal', 'xsd:float')
def _parse_aeolus_data(self, document, or_limit=None):
model = Model(self.graph)
rxcui_curie = "RXCUI:{}".format(document['aeolus']['rxcui'])
uni_curie = "UNII:{}".format(document['aeolus']['unii'])
model.addLabel(rxcui_curie, document['aeolus']['drug_name'])
model.addLabel(uni_curie, document['aeolus']['drug_name'])
model.addSameIndividual(rxcui_curie, uni_curie)
self.graph.addTriple(
rxcui_curie, self.globaltt['inchi_key'], document['unii']['inchikey'],
object_is_literal=True)
if or_limit is not None:
outcomes = (outcome for outcome in document['aeolus']['outcomes']
if 'ror' in outcome and outcome['ror'] >= or_limit)
else:
outcomes = (outcome for outcome in document['aeolus']['outcomes'])
for outcome in outcomes:
drug2outcome_assoc = Assoc(self.graph, self.name)
meddra_curie = "MEDDRA:{}".format(outcome['code'])
model.addLabel(meddra_curie, outcome['name'])
drug2outcome_assoc.sub = rxcui_curie
drug2outcome_assoc.obj = meddra_curie
drug2outcome_assoc.rel = self.globaltt['causes_or_contributes']
drug2outcome_assoc.description = \
"A proportional reporting ratio or odds " \
"ratio greater than or equal to {} in the " \
"AEOLUS data was the significance cut-off " \
"used for creating drug-outcome associations".format(or_limit)
drug2outcome_assoc.add_association_to_graph()
drug2outcome_assoc.add_predicate_object(
self.globaltt['probabalistic_quantifier'], outcome['ror'], 'Literal')
self._add_outcome_evidence(drug2outcome_assoc.assoc_id, outcome)
self._add_outcome_provenance(drug2outcome_assoc.assoc_id, outcome)
def _add_gene_anatomy_association(self, gene_id, anatomy_curie, rank):
"""
:param gene_id: str Non curified ID
:param gene_label: str Gene symbol
:param anatomy_curie: str curified anatomy term
:param rank: str rank
:return: None
"""
g2a_association = Assoc(self.graph, self.name)
genotype = Genotype(self.graph)
model = Model(self.graph)
gene_curie = "ENSEMBL:{}".format(gene_id)
rank = re.sub(r',', '', rank)
model.addIndividualToGraph(ind_id=gene_curie, label=None,
ind_type=genotype.genoparts['gene'])
g2a_association.sub = gene_curie
g2a_association.obj = anatomy_curie
g2a_association.rel = Assoc.object_properties['expressed_in']
g2a_association.add_association_to_graph()
g2a_association.add_predicate_object(
Assoc.datatype_properties['has_quantifier'],
float(rank), 'Literal', 'xsd:float')
return
def _add_gene_anatomy_association(self, gene_id, anatomy_curie, rank):
"""
:param gene_id: str Non curified ID
:param gene_label: str Gene symbol
:param anatomy_curie: str curified anatomy term
:param rank: str rank
:return: None
"""
g2a_association = Assoc(self.graph, self.name)
model = Model(self.graph)
gene_curie = "ENSEMBL:{}".format(gene_id)
rank = re.sub(r',', '', str(rank)) # ? can't do RE on a float ...
model.addIndividualToGraph(gene_curie, None)
g2a_association.sub = gene_curie
g2a_association.obj = anatomy_curie
g2a_association.rel = self.globaltt['expressed in']
g2a_association.add_association_to_graph()
g2a_association.add_predicate_object(
self.globaltt['has_quantifier'], float(rank), 'Literal', 'xsd:float')
return
def process_gaf(self, file, limit, id_map=None, eco_map=None):
if self.test_mode:
graph = self.testgraph
else:
graph = self.graph
model = Model(graph)
geno = Genotype(graph)
LOG.info("Processing Gene Associations from %s", file)
line_counter = 0
uniprot_hit = 0
uniprot_miss = 0
if '7955' in self.tax_ids:
zfin = ZFIN(self.graph_type, self.are_bnodes_skized)
if '6239' in self.tax_ids:
wbase = WormBase(self.graph_type, self.are_bnodes_skized)
with gzip.open(file, 'rb') as csvfile:
filereader = csv.reader(io.TextIOWrapper(csvfile, newline=""),
delimiter='\t',
quotechar='\"')
for row in filereader:
line_counter += 1
# comments start with exclamation
if re.match(r'!', ''.join(row)):
continue
if len(row) > 17 or len(row) < 15:
LOG.warning(
"Wrong number of columns %i, expected 15 or 17\n%s",
len(row), row)
continue
if 17 > len(row) >= 15:
row += [""] * (17 - len(row))
(dbase, gene_num, gene_symbol, qualifier, go_id, ref,
eco_symbol, with_or_from, aspect, gene_name, gene_synonym,
object_type, taxon, date, assigned_by, annotation_extension,
gene_product_form_id) = row
# test for required fields
if (dbase == '' or gene_num == '' or gene_symbol == ''
or go_id == '' or ref == '' or eco_symbol == ''
or aspect == '' or object_type == '' or taxon == ''
or date == '' or assigned_by == ''):
LOG.error(
"Missing required part of annotation on row %d:\n" +
'\t'.join(row), line_counter)
continue
# deal with qualifier NOT, contributes_to, colocalizes_with
if re.search(r'NOT', qualifier):
continue
if dbase in self.localtt:
dbase = self.localtt[dbase]
uniprotid = None
gene_id = None
if dbase == 'UniProtKB':
if id_map is not None and gene_num in id_map:
gene_id = id_map[gene_num]
uniprotid = ':'.join((dbase, gene_num))
(dbase, gene_num) = gene_id.split(':')
uniprot_hit += 1
else:
# LOG.warning(
# "UniProt id %s is without a 1:1 mapping to entrez/ensembl",
# gene_num)
uniprot_miss += 1
continue
else:
gene_num = gene_num.split(':')[-1] # last
gene_id = ':'.join((dbase, gene_num))
if self.test_mode and not (re.match(r'NCBIGene', gene_id)
and int(gene_num) in self.test_ids):
continue
model.addClassToGraph(gene_id, gene_symbol)
if gene_name != '':
model.addDescription(gene_id, gene_name)
if gene_synonym != '':
for syn in re.split(r'\|', gene_synonym):
model.addSynonym(gene_id, syn.strip())
if re.search(r'\|', taxon):
# TODO add annotations with >1 taxon
LOG.info(">1 taxon (%s) on line %d. skipping", taxon,
line_counter)
else:
tax_id = re.sub(r'taxon:', 'NCBITaxon:', taxon)
geno.addTaxon(tax_id, gene_id)
assoc = Assoc(graph, self.name)
assoc.set_subject(gene_id)
assoc.set_object(go_id)
try:
eco_id = eco_map[eco_symbol]
assoc.add_evidence(eco_id)
except KeyError:
LOG.error("Evidence code (%s) not mapped", eco_symbol)
refs = re.split(r'\|', ref)
for ref in refs:
ref = ref.strip()
if ref != '':
prefix = ref.split(':')[0] # sidestep 'MGI:MGI:'
if prefix in self.localtt:
prefix = self.localtt[prefix]
ref = ':'.join((prefix, ref.split(':')[-1]))
refg = Reference(graph, ref)
if prefix == 'PMID':
ref_type = self.globaltt['journal article']
refg.setType(ref_type)
refg.addRefToGraph()
assoc.add_source(ref)
# TODO add the source of the annotations from assigned by?
rel = self.resolve(aspect, mandatory=False)
if rel is not None and aspect == rel:
if aspect == 'F' and re.search(r'contributes_to',
qualifier):
assoc.set_relationship(self.globaltt['contributes to'])
else:
LOG.error(
"Aspect: %s with qualifier: %s is not recognized",
aspect, qualifier)
elif rel is not None:
assoc.set_relationship(rel)
assoc.add_association_to_graph()
else:
LOG.warning("No predicate for association \n%s\n",
str(assoc))
if uniprotid is not None:
assoc.set_description('Mapped from ' + uniprotid)
# object_type should be one of:
# protein_complex; protein; transcript; ncRNA; rRNA; tRNA;
# snRNA; snoRNA; any subtype of ncRNA in the Sequence Ontology.
# If the precise product type is unknown,
# gene_product should be used
#######################################################################
# Derive G2P Associations from IMP annotations
# in version 2.1 Pipe will indicate 'OR'
# and Comma will indicate 'AND'.
# in version 2.0, multiple values are separated by pipes
# where the pipe has been used to mean 'AND'
if eco_symbol == 'IMP' and with_or_from != '':
withitems = re.split(r'\|', with_or_from)
phenotypeid = go_id + 'PHENOTYPE'
# create phenotype associations
for i in withitems:
if i == '' or re.match(
r'(UniProtKB|WBPhenotype|InterPro|HGNC)', i):
LOG.warning(
"Don't know what having a uniprot id " +
"in the 'with' column means of %s", uniprotid)
continue
i = re.sub(r'MGI\:MGI\:', 'MGI:', i)
i = re.sub(r'WB:', 'WormBase:', i)
# for worms and fish, they might give a RNAi or MORPH
# in these cases make a reagent-targeted gene
if re.search('MRPHLNO|CRISPR|TALEN', i):
targeted_gene_id = zfin.make_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(
i, gene_id, targeted_gene_id)
# TODO PYLINT why is this needed?
# Redefinition of assoc type from
# dipper.models.assoc.Association.Assoc to
# dipper.models.assoc.G2PAssoc.G2PAssoc
assoc = G2PAssoc(graph, self.name,
targeted_gene_id, phenotypeid)
elif re.search(r'WBRNAi', i):
targeted_gene_id = wbase.make_reagent_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(
i, gene_id, targeted_gene_id)
assoc = G2PAssoc(graph, self.name,
targeted_gene_id, phenotypeid)
else:
assoc = G2PAssoc(graph, self.name, i, phenotypeid)
for ref in refs:
ref = ref.strip()
if ref != '':
prefix = ref.split(':')[0]
if prefix in self.localtt:
prefix = self.localtt[prefix]
ref = ':'.join((prefix, ref.split(':')[-1]))
assoc.add_source(ref)
# experimental phenotypic evidence
assoc.add_evidence(self.globaltt[
'experimental phenotypic evidence'])
assoc.add_association_to_graph()
# TODO should the G2PAssoc be
# the evidence for the GO assoc?
if not self.test_mode and limit is not None and line_counter > limit:
break
uniprot_tot = (uniprot_hit + uniprot_miss)
uniprot_per = 0.0
if uniprot_tot != 0:
uniprot_per = 100.0 * uniprot_hit / uniprot_tot
LOG.info(
"Uniprot: %.2f%% of %i benefited from the 1/4 day id mapping download",
uniprot_per, uniprot_tot)
return
def make_association(self, record):
"""
contstruct the association
:param record:
:return: modeled association of genotype to mammalian phenotype
"""
# prep record
# remove description and mapp Experiment Type to apo term
experiment_type = record['Experiment Type'].split('(')[0]
experiment_type = experiment_type.split(',')
record['experiment_type'] = list()
for exp_type in experiment_type:
exp_type = exp_type.lstrip().rstrip()
record['experiment_type'].append(
{
'id': self.apo_term_id[exp_type],
'term': exp_type,
})
sgd_phenotype = record['Phenotype']
pheno_obj = {
'entity': {
'term': None,
'apo_id': None
},
'quality': {
'term': None,
'apo_id': None
},
'has_quality': False # False = phenotype was descriptive and don't bother looking for a quality
}
phenotype = record['Phenotype']
if ':' in phenotype:
pheno_obj['has_quality'] = True
ent_qual = sgd_phenotype.split(': ')
entity = ent_qual[0]
quality = ent_qual[1]
pheno_obj['entity']['term'] = entity
pheno_obj['entity']['apo_id'] = self.apo_term_id[entity]
pheno_obj['quality']['term'] = quality
pheno_obj['quality']['apo_id'] = self.apo_term_id[quality]
else:
pheno_obj['entity']['term'] = phenotype
pheno_obj['entity']['apo_id'] = self.apo_term_id[phenotype]
record['pheno_obj'] = pheno_obj
# begin modeling
model = Model(self.graph)
# define the triple
gene = 'SGD:{}'.format(record['SGDID'])
relation = Model.object_properties['has_phenotype'] # has phenotype
if record['pheno_obj']['has_quality']:
pheno_label = '{0}:{1}'.format(
record['pheno_obj']['entity']['term'],
record['pheno_obj']['quality']['term'])
pheno_id = 'MONARCH:{0}{1}'.format(
record['pheno_obj']['entity']['apo_id'].replace(':', '_'),
record['pheno_obj']['quality']['apo_id'].replace(':', '_')
)
g2p_assoc = Assoc(self.graph, self.name, sub=gene, obj=pheno_id, pred=relation)
else:
pheno_label = record['pheno_obj']['entity']['term']
pheno_id = record['pheno_obj']['entity']['apo_id']
g2p_assoc = Assoc(self.graph, self.name, sub=gene, obj=pheno_id, pred=relation)
assoc_id = g2p_assoc.make_association_id(definedby='yeastgenome.org', subject=gene, predicate=relation,
object=pheno_id)
g2p_assoc.set_association_id(assoc_id=assoc_id)
# add to graph to mint assoc id
g2p_assoc.add_association_to_graph()
model.addLabel(subject_id=gene, label=record['Gene Name'])
# add the association triple
model.addTriple(subject_id=gene, predicate_id=relation, obj=pheno_id)
# make pheno subclass of UPHENO:0001001
model.addTriple(subject_id=pheno_id, predicate_id=Model.object_properties['subclass_of'], obj='UPHENO:0001001')
# label nodes
# pheno label
model.addLabel(subject_id=pheno_id, label=pheno_label)
g2p_assoc.description = self._make_description(record)
# add the references
references = record['Reference']
references = references.replace(' ', '')
references = references.split('|')
# created RGDRef prefix in curie map to route to proper reference URL in RGD
if len(references) > 0:
# make first ref in list the source
g2p_assoc.add_source(identifier=references[0])
ref_model = Reference(
self.graph, references[0],
Reference.ref_types['publication']
)
ref_model.addRefToGraph()
if len(references) > 1:
# create equivalent source for any other refs in list
for ref in references[1:]:
model.addSameIndividual(sub=references[0], obj=ref)
# add experiment type as evidence
for exp_type in record['experiment_type']:
g2p_assoc.add_evidence(exp_type['id'])
model.addLabel(subject_id=exp_type['id'], label=exp_type['term'])
try:
g2p_assoc.add_association_to_graph()
except Exception as e:
print(e)
return
def process_gaf(self, gaffile, limit, id_map=None, eco_map=None):
if self.test_mode:
graph = self.testgraph
else:
graph = self.graph
model = Model(graph)
geno = Genotype(graph)
LOG.info("Processing Gene Associations from %s", gaffile)
uniprot_hit = 0
uniprot_miss = 0
col = self.gaf_columns
with gzip.open(gaffile, 'rb') as csvfile:
reader = csv.reader(
io.TextIOWrapper(csvfile, newline=""), delimiter='\t', quotechar='\"')
for row in reader:
# comments start with exclamation
if row[0][0] == '!':
continue
if len(row) != len(col):
LOG.error(
"Wrong number of columns %i, expected ... got:\n\t%s",
len(col), row)
exit(1)
dbase = row[col.index('DB')].strip()
gene_num = row[col.index('DB_Object_ID')].strip()
gene_symbol = row[col.index('DB_Object_Symbol')].strip()
qualifier = row[col.index('Qualifier')]
go_id = row[col.index('GO_ID')].strip()
ref = row[col.index('DB:Reference')].strip()
eco_symbol = row[col.index('Evidence Code')].strip()
with_or_from = row[col.index('With (or) From')]
aspect = row[col.index('Aspect')].strip()
gene_name = row[col.index('DB_Object_Name')]
gene_synonym = row[col.index('DB_Object_Synonym')]
# object_type = row[col.index('DB_Object_Type')].strip()
taxon = row[col.index('Taxon and Interacting taxon')].strip()
# date = row[col.index('Date')].strip()
# assigned_by = row[col.index('Assigned_By')].strip()
# annotation_extension = row[col.index('Annotation_Extension')]
# gene_product_form_id = row[col.index('Gene_Product_Form_ID')]
# test for required fields
if '' in [row[:10], row[12]]:
LOG.error(
"Missing required part of annotation on row %i:\n%s",
reader.line_num, str(row[:-4]))
continue
# (Don't) deal with qualifier NOT, contributes_to, colocalizes_with
if re.search(r'NOT', qualifier):
continue
if dbase in self.localtt:
dbase = self.localtt[dbase]
uniprotid = None
gene_id = None
if dbase == 'UniProtKB':
if id_map is not None and gene_num in id_map:
gene_id = id_map[gene_num]
uniprotid = ':'.join((dbase, gene_num))
(dbase, gene_num) = gene_id.split(':')
uniprot_hit += 1
else:
# LOG.warning(
# "UniProt id %s is without a 1:1 mapping to entrez/ensembl",
# gene_num)
uniprot_miss += 1
continue
else:
gene_num = gene_num.split(':')[-1] # last
gene_id = ':'.join((dbase, gene_num))
if self.test_mode and gene_id[:9] != 'NCBIGene:' and\
gene_num not in self.test_ids:
continue
model.addClassToGraph(gene_id, gene_symbol)
if gene_name != '':
model.addDescription(gene_id, gene_name)
if gene_synonym != '':
for syn in re.split(r'\|', gene_synonym):
syn = syn.strip()
if syn[:10] == 'UniProtKB:':
model.addTriple(
gene_id, self.globaltt['has gene product'], syn)
elif re.fullmatch(graph.curie_regexp, syn) is not None:
LOG.warning(
'possible curie "%s" as a literal synomym for %s',
syn, gene_id)
model.addSynonym(gene_id, syn)
else:
model.addSynonym(gene_id, syn)
for txid in taxon.split('|'):
tax_curie = re.sub(r'taxon:', 'NCBITaxon:', txid)
geno.addTaxon(tax_curie, gene_id)
assoc = Assoc(graph, self.name)
assoc.set_subject(gene_id)
assoc.set_object(go_id)
try:
eco_id = eco_map[eco_symbol]
assoc.add_evidence(eco_id)
except KeyError:
LOG.error("Evidence code (%s) not mapped", eco_symbol)
refs = re.split(r'\|', ref)
for ref in refs:
ref = ref.strip()
if ref != '':
prefix = ref.split(':')[0] # sidestep 'MGI:MGI:'
if prefix in self.localtt:
prefix = self.localtt[prefix]
ref = ':'.join((prefix, ref.split(':')[-1]))
refg = Reference(graph, ref)
if prefix == 'PMID':
ref_type = self.globaltt['journal article']
refg.setType(ref_type)
refg.addRefToGraph()
assoc.add_source(ref)
# TODO add the source of the annotations from assigned by?
rel = self.resolve(aspect, mandatory=False)
if rel is not None and aspect == rel:
if aspect == 'F' and re.search(r'contributes_to', qualifier):
assoc.set_relationship(self.globaltt['contributes to'])
else:
LOG.error(
"Aspect: %s with qualifier: %s is not recognized",
aspect, qualifier)
elif rel is not None:
assoc.set_relationship(rel)
assoc.add_association_to_graph()
else:
LOG.warning("No predicate for association \n%s\n", str(assoc))
if uniprotid is not None:
assoc.set_description('Mapped from ' + uniprotid)
# object_type should be one of:
# protein_complex; protein; transcript; ncRNA; rRNA; tRNA;
# snRNA; snoRNA; any subtype of ncRNA in the Sequence Ontology.
# If the precise product type is unknown,
# gene_product should be used
########################################################################
# Derive G2P Associations from IMP annotations
# in version 2.1 Pipe will indicate 'OR'
# and Comma will indicate 'AND'.
# in version 2.0, multiple values are separated by pipes
# where the pipe has been used to mean 'AND'
if eco_symbol == 'IMP' and with_or_from != '':
withitems = with_or_from.split('|')
phenotypeid = go_id + 'PHENOTYPE'
# create phenotype associations
for itm in withitems:
if itm == '' or re.match(
r'(UniProtKB|WBPhenotype|InterPro|HGNC)', itm):
LOG.warning(
"Skipping %s from or with %s", uniprotid, itm)
continue
itm = re.sub(r'MGI\:MGI\:', 'MGI:', itm)
itm = re.sub(r'WB:', 'WormBase:', itm)
# for worms and fish, they might give a RNAi or MORPH
# in these cases make a reagent-targeted gene
if re.search('MRPHLNO|CRISPR|TALEN', itm):
targeted_gene_id = self.zfin.make_targeted_gene_id(
gene_id, itm)
geno.addReagentTargetedGene(itm, gene_id, targeted_gene_id)
# TODO PYLINT why is this needed?
# Redefinition of assoc type from
# dipper.models.assoc.Association.Assoc to
# dipper.models.assoc.G2PAssoc.G2PAssoc
assoc = G2PAssoc(
graph, self.name, targeted_gene_id, phenotypeid)
elif re.search(r'WBRNAi', itm):
targeted_gene_id = self.wbase.make_reagent_targeted_gene_id(
gene_id, itm)
geno.addReagentTargetedGene(itm, gene_id, targeted_gene_id)
assoc = G2PAssoc(
graph, self.name, targeted_gene_id, phenotypeid)
else:
assoc = G2PAssoc(graph, self.name, itm, phenotypeid)
for ref in refs:
ref = ref.strip()
if ref != '':
prefix = ref.split(':')[0]
if prefix in self.localtt:
prefix = self.localtt[prefix]
ref = ':'.join((prefix, ref.split(':')[-1]))
assoc.add_source(ref)
# experimental phenotypic evidence
assoc.add_evidence(
self.globaltt['experimental phenotypic evidence'])
assoc.add_association_to_graph()
# TODO should the G2PAssoc be the evidence for the GO assoc?
if not self.test_mode and limit is not None and \
reader.line_num > limit:
break
uniprot_tot = (uniprot_hit + uniprot_miss)
uniprot_per = 0.0
if uniprot_tot != 0:
uniprot_per = 100.0 * uniprot_hit / uniprot_tot
LOG.info(
"Uniprot: %.2f%% of %i benefited from the 1/4 day id mapping download",
uniprot_per, uniprot_tot)
def make_association(self, record):
"""
contstruct the association
:param record:
:return: modeled association of genotype to mammalian??? phenotype
"""
# prep record
# remove description and mapp Experiment Type to apo term
experiment_type = record['Experiment Type'].split('(')[0]
experiment_type = experiment_type.split(',')
record['experiment_type'] = list()
for exp_type in experiment_type:
exp_type = exp_type.lstrip().rstrip()
record['experiment_type'].append({
'id': self.apo_term_id[exp_type],
'term': exp_type,
})
sgd_phenotype = record['Phenotype']
pheno_obj = {
'entity': {
'term': None,
'apo_id': None
},
'quality': {
'term': None,
'apo_id': None
},
'has_quality':
False # descriptive and don't bother looking for a quality
}
phenotype = record['Phenotype']
if ':' in phenotype:
pheno_obj['has_quality'] = True
ent_qual = sgd_phenotype.split(': ')
entity = ent_qual[0]
quality = ent_qual[1]
pheno_obj['entity']['term'] = entity
pheno_obj['entity']['apo_id'] = self.apo_term_id[entity]
pheno_obj['quality']['term'] = quality
pheno_obj['quality']['apo_id'] = self.apo_term_id[quality]
else:
pheno_obj['entity']['term'] = phenotype
pheno_obj['entity']['apo_id'] = self.apo_term_id[phenotype]
record['pheno_obj'] = pheno_obj
# begin modeling
model = Model(self.graph)
# define the triple
gene = 'SGD:{}'.format(record['SGDID'])
relation = self.globaltt['has phenotype']
if record['pheno_obj']['has_quality']:
pheno_label = '{0}:{1}'.format(
record['pheno_obj']['entity']['term'],
record['pheno_obj']['quality']['term'])
pheno_id = 'MONARCH:{0}{1}'.format(
record['pheno_obj']['entity']['apo_id'].replace(':', '_'),
record['pheno_obj']['quality']['apo_id'].replace(':', '_'))
g2p_assoc = Assoc(self.graph,
self.name,
sub=gene,
obj=pheno_id,
pred=relation)
else:
pheno_label = record['pheno_obj']['entity']['term']
pheno_id = record['pheno_obj']['entity']['apo_id']
g2p_assoc = Assoc(self.graph,
self.name,
sub=gene,
obj=pheno_id,
pred=relation)
assoc_id = g2p_assoc.make_association_id('yeastgenome.org', gene,
relation, pheno_id)
g2p_assoc.set_association_id(assoc_id=assoc_id)
# add to graph to mint assoc id
g2p_assoc.add_association_to_graph()
model.addLabel(subject_id=gene, label=record['Gene Name'])
# add the association triple
model.addTriple(subject_id=gene, predicate_id=relation, obj=pheno_id)
model.addTriple(subject_id=pheno_id,
predicate_id=self.globaltt['subclass_of'],
obj=self.globaltt['Phenotype'])
# label nodes
# pheno label
model.addLabel(subject_id=pheno_id, label=pheno_label)
g2p_assoc.description = self._make_description(record)
# add the references
references = record['Reference']
references = references.replace(' ', '')
references = references.split('|')
# created Ref prefix in curie map to route to proper reference URL in SGD
if len(references) > 0:
# make first ref in list the source
g2p_assoc.add_source(identifier=references[0])
ref_model = Reference(self.graph, references[0],
self.globaltt['publication'])
ref_model.addRefToGraph()
if len(references) > 1:
# create equivalent source for any other refs in list
for ref in references[1:]:
model.addSameIndividual(sub=references[0], obj=ref)
# add experiment type as evidence
for exp_type in record['experiment_type']:
g2p_assoc.add_evidence(exp_type['id'])
model.addLabel(subject_id=exp_type['id'], label=exp_type['term'])
try:
g2p_assoc.add_association_to_graph()
except Exception as e:
print(e)
return
def _add_therapy_drug_association(self, drug_id, disease_id, therapy_status_id):
"""
Create an association linking a drug and disease with
RO:0002606 (substance_that_treats) and any supporting information
such as FDA approval and source (not implemented)
:param drug_id: Id as curie of the drug
:param disease_id: Id as curie of the disease
:param therapy_status: (Optional) String label of therapy approval status
:return: None
"""
gu = GraphUtils(curie_map.get())
# Placeholder relationship, note this does not exist in RO
relationship_id = "RO:has_approval_status"
gu.addTriple(self.graph, drug_id, gu.object_properties['substance_that_treats'], disease_id)
# Make association
drug_disease_annot = self.make_cgd_id("assoc{0}{1}".format(drug_id, disease_id))
therapy_disease_assoc = Assoc(self.name)
therapy_disease_assoc.set_subject(drug_id)
therapy_disease_assoc.set_relationship(gu.object_properties['substance_that_treats'])
therapy_disease_assoc.set_object(disease_id)
therapy_disease_assoc.set_association_id(drug_disease_annot)
therapy_disease_assoc.add_association_to_graph(self.graph)
gu.addTriple(self.graph, drug_disease_annot, relationship_id, therapy_status_id)
def make_association(self, record):
"""
contstruct the association
:param record:
:return: modeled association of genotype to mammalian phenotype
"""
# prep record
# remove description and mapp Experiment Type to apo term
experiment_type = record['Experiment Type'].split('(')[0]
experiment_type = experiment_type.split(',')
record['experiment_type'] = list()
for exp_type in experiment_type:
exp_type = exp_type.lstrip().rstrip()
record['experiment_type'].append({
'id': self.apo_term_id[exp_type],
'term': exp_type,
})
sgd_phenotype = record['Phenotype']
pheno_obj = {
'entity': {
'term': None,
'apo_id': None
},
'quality': {
'term': None,
'apo_id': None
},
'has_quality':
False # False = phenotype was descriptive and don't bother looking for a quality
}
phenotype = record['Phenotype']
if ':' in phenotype:
pheno_obj['has_quality'] = True
ent_qual = sgd_phenotype.split(': ')
entity = ent_qual[0]
quality = ent_qual[1]
pheno_obj['entity']['term'] = entity
pheno_obj['entity']['apo_id'] = self.apo_term_id[entity]
pheno_obj['quality']['term'] = quality
pheno_obj['quality']['apo_id'] = self.apo_term_id[quality]
else:
pheno_obj['entity']['term'] = phenotype
pheno_obj['entity']['apo_id'] = self.apo_term_id[phenotype]
record['pheno_obj'] = pheno_obj
# begin modeling
model = Model(self.graph)
# define the triple
gene = 'SGD:{}'.format(record['SGDID'])
relation = Model.object_properties['has_phenotype'] # has phenotype
if record['pheno_obj']['has_quality']:
pheno_label = '{0}:{1}'.format(
record['pheno_obj']['entity']['term'],
record['pheno_obj']['quality']['term'])
pheno_id = 'MONARCH:{0}{1}'.format(
record['pheno_obj']['entity']['apo_id'].replace(':', '_'),
record['pheno_obj']['quality']['apo_id'].replace(':', '_'))
g2p_assoc = Assoc(self.graph,
self.name,
sub=gene,
obj=pheno_id,
pred=relation)
else:
pheno_label = record['pheno_obj']['entity']['term']
pheno_id = record['pheno_obj']['entity']['apo_id']
g2p_assoc = Assoc(self.graph,
self.name,
sub=gene,
obj=pheno_id,
pred=relation)
assoc_id = g2p_assoc.make_association_id(
definedby='yeastgenome.org',
subject=gene,
predicate=relation,
object=pheno_id)
g2p_assoc.set_association_id(assoc_id=assoc_id)
# add to graph to mint assoc id
g2p_assoc.add_association_to_graph()
model.addLabel(subject_id=gene, label=record['Gene Name'])
# add the association triple
model.addTriple(subject_id=gene, predicate_id=relation, obj=pheno_id)
# make pheno subclass of UPHENO:0001001
model.addTriple(subject_id=pheno_id,
predicate_id=Model.object_properties['subclass_of'],
obj='UPHENO:0001001')
# label nodes
# pheno label
model.addLabel(subject_id=pheno_id, label=pheno_label)
# add the descripiton: all the unmodeled data in a '|' delimited list
description = [
'genomic_background: {}'.format(record['Strain Background']),
'allele: {}'.format(record['Allele']),
'chemical: {}'.format(record['Chemical']),
'condition: {}'.format(record['Condition']),
'details: {}'.format(record['Details']),
'feature_name: {}'.format(record['Feature Name']),
'gene_name: {}'.format(record['Gene Name']),
'mutant_type: {}'.format(record['Mutant Type']),
'reporter: {}'.format(record['Reporter']),
]
g2p_assoc.description = " | ".join(description)
# add the references
references = record['Reference']
references = references.replace(' ', '')
references = references.split('|')
# created RGDRef prefix in curie map to route to proper reference URL in RGD
if len(references) > 0:
# make first ref in list the source
g2p_assoc.add_source(identifier=references[0])
ref_model = Reference(self.graph, references[0],
Reference.ref_types['publication'])
ref_model.addRefToGraph()
if len(references) > 1:
# create equivalent source for any other refs in list
for ref in references[1:]:
model.addSameIndividual(sub=references[0], obj=ref)
# add experiment type as evidence
for exp_type in record['experiment_type']:
g2p_assoc.add_evidence(exp_type['id'])
model.addLabel(subject_id=exp_type['id'], label=exp_type['term'])
try:
g2p_assoc.add_association_to_graph()
except Exception as e:
print(e)
return
def process_gaf(self, file, limit, id_map=None, eco_map=None):
if self.test_mode:
graph = self.testgraph
else:
graph = self.graph
model = Model(graph)
geno = Genotype(graph)
LOG.info("Processing Gene Associations from %s", file)
line_counter = 0
uniprot_hit = 0
uniprot_miss = 0
if 7955 in self.tax_ids:
zfin = ZFIN(self.graph_type, self.are_bnodes_skized)
if 6239 in self.tax_ids:
wbase = WormBase(self.graph_type, self.are_bnodes_skized)
with gzip.open(file, 'rb') as csvfile:
filereader = csv.reader(
io.TextIOWrapper(csvfile, newline=""), delimiter='\t', quotechar='\"')
for row in filereader:
line_counter += 1
# comments start with exclamation
if re.match(r'!', ''.join(row)):
continue
if len(row) > 17 or len(row) < 15:
LOG.warning(
"Wrong number of columns %i, expected 15 or 17\n%s",
len(row), row)
continue
if 17 > len(row) >= 15:
row += [""] * (17 - len(row))
(dbase,
gene_num,
gene_symbol,
qualifier,
go_id,
ref,
eco_symbol,
with_or_from,
aspect,
gene_name,
gene_synonym,
object_type,
taxon,
date,
assigned_by,
annotation_extension,
gene_product_form_id) = row
# test for required fields
if (dbase == '' or gene_num == '' or gene_symbol == '' or
go_id == '' or ref == '' or eco_symbol == '' or
aspect == '' or object_type == '' or taxon == '' or
date == '' or assigned_by == ''):
LOG.error(
"Missing required part of annotation on row %d:\n"+'\t'
.join(row), line_counter)
continue
# deal with qualifier NOT, contributes_to, colocalizes_with
if re.search(r'NOT', qualifier):
continue
if dbase in self.localtt:
dbase = self.localtt[dbase]
uniprotid = None
gene_id = None
if dbase == 'UniProtKB':
if id_map is not None and gene_num in id_map:
gene_id = id_map[gene_num]
uniprotid = ':'.join((dbase, gene_num))
(dbase, gene_num) = gene_id.split(':')
uniprot_hit += 1
else:
# LOG.warning(
# "UniProt id %s is without a 1:1 mapping to entrez/ensembl",
# gene_num)
uniprot_miss += 1
continue
else:
gene_num = gene_num.split(':')[-1] # last
gene_id = ':'.join((dbase, gene_num))
if self.test_mode and not(
re.match(r'NCBIGene', gene_id) and
int(gene_num) in self.test_ids):
continue
model.addClassToGraph(gene_id, gene_symbol)
if gene_name != '':
model.addDescription(gene_id, gene_name)
if gene_synonym != '':
for syn in re.split(r'\|', gene_synonym):
model.addSynonym(gene_id, syn.strip())
if re.search(r'\|', taxon):
# TODO add annotations with >1 taxon
LOG.info(
">1 taxon (%s) on line %d. skipping", taxon, line_counter)
else:
tax_id = re.sub(r'taxon:', 'NCBITaxon:', taxon)
geno.addTaxon(tax_id, gene_id)
assoc = Assoc(graph, self.name)
assoc.set_subject(gene_id)
assoc.set_object(go_id)
try:
eco_id = eco_map[eco_symbol]
assoc.add_evidence(eco_id)
except KeyError:
LOG.error("Evidence code (%s) not mapped", eco_symbol)
refs = re.split(r'\|', ref)
for ref in refs:
ref = ref.strip()
if ref != '':
prefix = ref.split(':')[0] # sidestep 'MGI:MGI:'
if prefix in self.localtt:
prefix = self.localtt[prefix]
ref = ':'.join((prefix, ref.split(':')[-1]))
refg = Reference(graph, ref)
if prefix == 'PMID':
ref_type = self.globaltt['journal article']
refg.setType(ref_type)
refg.addRefToGraph()
assoc.add_source(ref)
# TODO add the source of the annotations from assigned by?
rel = self.resolve(aspect, mandatory=False)
if rel is not None and aspect == rel:
if aspect == 'F' and re.search(r'contributes_to', qualifier):
assoc.set_relationship(self.globaltt['contributes to'])
else:
LOG.error(
"Aspect: %s with qualifier: %s is not recognized",
aspect, qualifier)
elif rel is not None:
assoc.set_relationship(rel)
assoc.add_association_to_graph()
else:
LOG.warning("No predicate for association \n%s\n", str(assoc))
if uniprotid is not None:
assoc.set_description('Mapped from ' + uniprotid)
# object_type should be one of:
# protein_complex; protein; transcript; ncRNA; rRNA; tRNA;
# snRNA; snoRNA; any subtype of ncRNA in the Sequence Ontology.
# If the precise product type is unknown,
# gene_product should be used
#######################################################################
# Derive G2P Associations from IMP annotations
# in version 2.1 Pipe will indicate 'OR'
# and Comma will indicate 'AND'.
# in version 2.0, multiple values are separated by pipes
# where the pipe has been used to mean 'AND'
if eco_symbol == 'IMP' and with_or_from != '':
withitems = re.split(r'\|', with_or_from)
phenotypeid = go_id+'PHENOTYPE'
# create phenotype associations
for i in withitems:
if i == '' or re.match(
r'(UniProtKB|WBPhenotype|InterPro|HGNC)', i):
LOG.warning(
"Don't know what having a uniprot id " +
"in the 'with' column means of %s", uniprotid)
continue
i = re.sub(r'MGI\:MGI\:', 'MGI:', i)
i = re.sub(r'WB:', 'WormBase:', i)
# for worms and fish, they might give a RNAi or MORPH
# in these cases make a reagent-targeted gene
if re.search('MRPHLNO|CRISPR|TALEN', i):
targeted_gene_id = zfin.make_targeted_gene_id(gene_id, i)
geno.addReagentTargetedGene(i, gene_id, targeted_gene_id)
# TODO PYLINT why is this needed?
# Redefinition of assoc type from
# dipper.models.assoc.Association.Assoc to
# dipper.models.assoc.G2PAssoc.G2PAssoc
assoc = G2PAssoc(
graph, self.name, targeted_gene_id, phenotypeid)
elif re.search(r'WBRNAi', i):
targeted_gene_id = wbase.make_reagent_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(i, gene_id, targeted_gene_id)
assoc = G2PAssoc(
graph, self.name, targeted_gene_id, phenotypeid)
else:
assoc = G2PAssoc(graph, self.name, i, phenotypeid)
for ref in refs:
ref = ref.strip()
if ref != '':
prefix = ref.split(':')[0]
if prefix in self.localtt:
prefix = self.localtt[prefix]
ref = ':'.join((prefix, ref.split(':')[-1]))
assoc.add_source(ref)
# experimental phenotypic evidence
assoc.add_evidence(
self.globaltt['experimental phenotypic evidence'])
assoc.add_association_to_graph()
# TODO should the G2PAssoc be
# the evidence for the GO assoc?
if not self.test_mode and limit is not None and line_counter > limit:
break
uniprot_tot = (uniprot_hit + uniprot_miss)
uniprot_per = 0.0
if uniprot_tot != 0:
uniprot_per = 100.0 * uniprot_hit / uniprot_tot
LOG.info(
"Uniprot: %f.2%% of %i benifited from the 1/4 day id mapping download",
uniprot_per, uniprot_tot)
return
def _add_component_pathway_association(self, gene_curie, pathway_curie,
pathway_label, eco_curie):
pathway = Pathway(self.graph)
pathway.addPathway(pathway_curie, pathway_label)
pathway.addComponentToPathway(gene_curie, pathway_curie)
association = Assoc(self.graph, self.name)
association.sub = gene_curie
association.rel = self.globaltt['involved in']
association.obj = pathway_curie
association.set_association_id()
association.add_evidence(eco_curie)
association.add_association_to_graph()
def process_gaf(self, file, limit, id_map=None):
if self.testMode:
g = self.testgraph
else:
g = self.graph
model = Model(g)
geno = Genotype(g)
logger.info("Processing Gene Associations from %s", file)
line_counter = 0
if 7955 in self.tax_ids:
zfin = ZFIN(self.graph_type, self.are_bnodes_skized)
elif 6239 in self.tax_ids:
wbase = WormBase(self.graph_type, self.are_bnodes_skized)
with gzip.open(file, 'rb') as csvfile:
filereader = csv.reader(io.TextIOWrapper(csvfile, newline=""),
delimiter='\t', quotechar='\"')
for row in filereader:
line_counter += 1
# comments start with exclamation
if re.match(r'!', ''.join(row)):
continue
(db, gene_num, gene_symbol, qualifier, go_id, ref, eco_symbol,
with_or_from, aspect, gene_name, gene_synonym, object_type,
taxon, date, assigned_by, annotation_extension,
gene_product_form_id) = row
# test for required fields
if (db == '' or gene_num == '' or gene_symbol == '' or
go_id == '' or ref == '' or eco_symbol == '' or
aspect == '' or object_type == '' or taxon == '' or
date == '' or assigned_by == ''):
logger.error(
"Missing required part of annotation " +
"on row %d:\n"+'\t'.join(row),
line_counter)
continue
# deal with qualifier NOT, contributes_to, colocalizes_with
if re.search(r'NOT', qualifier):
continue
db = self.clean_db_prefix(db)
uniprotid = None
gene_id = None
if db == 'UniProtKB':
mapped_ids = id_map.get(gene_num)
if id_map is not None and mapped_ids is not None:
if len(mapped_ids) == 1:
gene_id = mapped_ids[0]
uniprotid = ':'.join((db, gene_num))
gene_num = re.sub(r'\w+\:', '', gene_id)
elif len(mapped_ids) > 1:
# logger.warning(
# "Skipping gene id mapped for >1 gene %s -> %s",
# gene_num, str(mapped_ids))
continue
else:
continue
elif db == 'MGI':
gene_num = re.sub(r'MGI:', '', gene_num)
gene_id = ':'.join((db, gene_num))
gene_id = re.sub(r'MGI\:MGI\:', 'MGI:', gene_id)
else:
gene_id = ':'.join((db, gene_num))
if self.testMode \
and not(
re.match(r'NCBIGene', gene_id) and
int(gene_num) in self.test_ids):
continue
model.addClassToGraph(gene_id, gene_symbol)
if gene_name != '':
model.addDescription(gene_id, gene_name)
if gene_synonym != '':
for s in re.split(r'\|', gene_synonym):
model.addSynonym(gene_id, s.strip())
if re.search(r'\|', taxon):
# TODO add annotations with >1 taxon
logger.info(">1 taxon (%s) on line %d. skipping", taxon,
line_counter)
else:
tax_id = re.sub(r'taxon:', 'NCBITaxon:', taxon)
geno.addTaxon(tax_id, gene_id)
assoc = Assoc(g, self.name)
assoc.set_subject(gene_id)
assoc.set_object(go_id)
eco_id = self.map_go_evidence_code_to_eco(eco_symbol)
if eco_id is not None:
assoc.add_evidence(eco_id)
refs = re.split(r'\|', ref)
for r in refs:
r = r.strip()
if r != '':
prefix = re.split(r':', r)[0]
r = re.sub(prefix, self.clean_db_prefix(prefix), r)
r = re.sub(r'MGI\:MGI\:', 'MGI:', r)
ref = Reference(g, r)
if re.match(r'PMID', r):
ref_type = Reference.ref_types['journal_article']
ref.setType(ref_type)
ref.addRefToGraph()
assoc.add_source(r)
# TODO add the source of the annotations from assigned by?
aspect_rel_map = {
'P': model.object_properties['involved_in'], # involved in
'F': model.object_properties['enables'], # enables
'C': model.object_properties['part_of'] # part of
}
if aspect not in aspect_rel_map:
logger.error("Aspect not recognized: %s", aspect)
rel = aspect_rel_map.get(aspect)
if aspect == 'F' and re.search(r'contributes_to', qualifier):
rel = model.object_properties['contributes_to']
assoc.set_relationship(rel)
if uniprotid is not None:
assoc.set_description('Mapped from '+uniprotid)
# object_type should be one of:
# protein_complex; protein; transcript; ncRNA; rRNA; tRNA;
# snRNA; snoRNA; any subtype of ncRNA in the Sequence Ontology.
# If the precise product type is unknown,
# gene_product should be used
assoc.add_association_to_graph()
# Derive G2P Associations from IMP annotations
# in version 2.1 Pipe will indicate 'OR'
# and Comma will indicate 'AND'.
# in version 2.0, multiple values are separated by pipes
# where the pipe has been used to mean 'AND'
if eco_symbol == 'IMP' and with_or_from != '':
withitems = re.split(r'\|', with_or_from)
phenotypeid = go_id+'PHENOTYPE'
# create phenotype associations
for i in withitems:
if i == '' or \
re.match(
r'(UniProtKB|WBPhenotype|InterPro|HGNC)',
i):
logger.warning(
"Don't know what having a uniprot id " +
"in the 'with' column means of %s",
uniprotid)
continue
i = re.sub(r'MGI\:MGI\:', 'MGI:', i)
i = re.sub(r'WB:', 'WormBase:', i)
# for worms and fish, they might give a RNAi or MORPH
# in these cases make a reagent-targeted gene
if re.search('MRPHLNO|CRISPR|TALEN', i):
targeted_gene_id = zfin.make_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(i, gene_id,
targeted_gene_id)
# TODO PYLINT why is this:
# Redefinition of assoc type from
# dipper.models.assoc.Association.Assoc to
# dipper.models.assoc.G2PAssoc.G2PAssoc
assoc = G2PAssoc(g, self.name, targeted_gene_id,
phenotypeid)
elif re.search(r'WBRNAi', i):
targeted_gene_id = \
wbase.make_reagent_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(
i, gene_id, targeted_gene_id)
assoc = G2PAssoc(
g, self.name, targeted_gene_id, phenotypeid)
else:
assoc = G2PAssoc(g, self.name, i, phenotypeid)
for r in refs:
r = r.strip()
if r != '':
prefix = re.split(r':', r)[0]
r = re.sub(
prefix, self.clean_db_prefix(prefix), r)
r = re.sub(r'MGI\:MGI\:', 'MGI:', r)
assoc.add_source(r)
# experimental phenotypic evidence
assoc.add_evidence("ECO:0000059")
assoc.add_association_to_graph()
# TODO should the G2PAssoc be
# the evidence for the GO assoc?
if not self.testMode and \
limit is not None and line_counter > limit:
break
return
def _add_component_pathway_association(self, eco_map, component,
component_prefix, pathway_id,
pathway_prefix, pathway_label,
go_ecode):
pathway = Pathway(self.graph)
pathway_curie = "{}:{}".format(pathway_prefix, pathway_id)
gene_curie = "{}:{}".format(component_prefix, component.strip())
eco_curie = eco_map[go_ecode]
pathway.addPathway(pathway_curie, pathway_label)
pathway.addComponentToPathway(gene_curie, pathway_curie)
association = Assoc(self.graph, self.name)
association.sub = gene_curie
association.rel = self.globaltt['involved in']
association.obj = pathway_curie
association.set_association_id()
association.add_evidence(eco_curie)
association.add_association_to_graph()
return
def process_gaf(self, file, limit, id_map=None):
if self.testMode:
g = self.testgraph
else:
g = self.graph
model = Model(g)
geno = Genotype(g)
logger.info("Processing Gene Associations from %s", file)
line_counter = 0
if 7955 in self.tax_ids:
zfin = ZFIN(self.graph_type, self.are_bnodes_skized)
elif 6239 in self.tax_ids:
wbase = WormBase(self.graph_type, self.are_bnodes_skized)
with gzip.open(file, 'rb') as csvfile:
filereader = csv.reader(io.TextIOWrapper(csvfile, newline=""),
delimiter='\t',
quotechar='\"')
for row in filereader:
line_counter += 1
# comments start with exclamation
if re.match(r'!', ''.join(row)):
continue
(db, gene_num, gene_symbol, qualifier, go_id, ref, eco_symbol,
with_or_from, aspect, gene_name, gene_synonym, object_type,
taxon, date, assigned_by, annotation_extension,
gene_product_form_id) = row
# test for required fields
if (db == '' or gene_num == '' or gene_symbol == ''
or go_id == '' or ref == '' or eco_symbol == ''
or aspect == '' or object_type == '' or taxon == ''
or date == '' or assigned_by == ''):
logger.error(
"Missing required part of annotation " +
"on row %d:\n" + '\t'.join(row), line_counter)
continue
# deal with qualifier NOT, contributes_to, colocalizes_with
if re.search(r'NOT', qualifier):
continue
db = self.clean_db_prefix(db)
uniprotid = None
gene_id = None
if db == 'UniProtKB':
mapped_ids = id_map.get(gene_num)
if id_map is not None and mapped_ids is not None:
if len(mapped_ids) == 1:
gene_id = mapped_ids[0]
uniprotid = ':'.join((db, gene_num))
gene_num = re.sub(r'\w+\:', '', gene_id)
elif len(mapped_ids) > 1:
# logger.warning(
# "Skipping gene id mapped for >1 gene %s -> %s",
# gene_num, str(mapped_ids))
continue
else:
continue
elif db == 'MGI':
gene_num = re.sub(r'MGI:', '', gene_num)
gene_id = ':'.join((db, gene_num))
gene_id = re.sub(r'MGI\:MGI\:', 'MGI:', gene_id)
else:
gene_id = ':'.join((db, gene_num))
if self.testMode \
and not(
re.match(r'NCBIGene', gene_id) and
int(gene_num) in self.test_ids):
continue
model.addClassToGraph(gene_id, gene_symbol)
if gene_name != '':
model.addDescription(gene_id, gene_name)
if gene_synonym != '':
for s in re.split(r'\|', gene_synonym):
model.addSynonym(gene_id, s.strip())
if re.search(r'\|', taxon):
# TODO add annotations with >1 taxon
logger.info(">1 taxon (%s) on line %d. skipping", taxon,
line_counter)
else:
tax_id = re.sub(r'taxon:', 'NCBITaxon:', taxon)
geno.addTaxon(tax_id, gene_id)
assoc = Assoc(g, self.name)
assoc.set_subject(gene_id)
assoc.set_object(go_id)
eco_id = self.map_go_evidence_code_to_eco(eco_symbol)
if eco_id is not None:
assoc.add_evidence(eco_id)
refs = re.split(r'\|', ref)
for r in refs:
r = r.strip()
if r != '':
prefix = re.split(r':', r)[0]
r = re.sub(prefix, self.clean_db_prefix(prefix), r)
r = re.sub(r'MGI\:MGI\:', 'MGI:', r)
ref = Reference(g, r)
if re.match(r'PMID', r):
ref_type = Reference.ref_types['journal_article']
ref.setType(ref_type)
ref.addRefToGraph()
assoc.add_source(r)
# TODO add the source of the annotations from assigned by?
aspect_rel_map = {
'P': model.object_properties['involved_in'], # involved in
'F': model.object_properties['enables'], # enables
'C': model.object_properties['part_of'] # part of
}
if aspect not in aspect_rel_map:
logger.error("Aspect not recognized: %s", aspect)
rel = aspect_rel_map.get(aspect)
if aspect == 'F' and re.search(r'contributes_to', qualifier):
rel = model.object_properties['contributes_to']
assoc.set_relationship(rel)
if uniprotid is not None:
assoc.set_description('Mapped from ' + uniprotid)
# object_type should be one of:
# protein_complex; protein; transcript; ncRNA; rRNA; tRNA;
# snRNA; snoRNA; any subtype of ncRNA in the Sequence Ontology.
# If the precise product type is unknown,
# gene_product should be used
assoc.add_association_to_graph()
# Derive G2P Associations from IMP annotations
# in version 2.1 Pipe will indicate 'OR'
# and Comma will indicate 'AND'.
# in version 2.0, multiple values are separated by pipes
# where the pipe has been used to mean 'AND'
if eco_symbol == 'IMP' and with_or_from != '':
withitems = re.split(r'\|', with_or_from)
phenotypeid = go_id + 'PHENOTYPE'
# create phenotype associations
for i in withitems:
if i == '' or \
re.match(
r'(UniProtKB|WBPhenotype|InterPro|HGNC)',
i):
logger.warning(
"Don't know what having a uniprot id " +
"in the 'with' column means of %s", uniprotid)
continue
i = re.sub(r'MGI\:MGI\:', 'MGI:', i)
i = re.sub(r'WB:', 'WormBase:', i)
# for worms and fish, they might give a RNAi or MORPH
# in these cases make a reagent-targeted gene
if re.search('MRPHLNO|CRISPR|TALEN', i):
targeted_gene_id = zfin.make_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(
i, gene_id, targeted_gene_id)
# TODO PYLINT why is this:
# Redefinition of assoc type from
# dipper.models.assoc.Association.Assoc to
# dipper.models.assoc.G2PAssoc.G2PAssoc
assoc = G2PAssoc(g, self.name, targeted_gene_id,
phenotypeid)
elif re.search(r'WBRNAi', i):
targeted_gene_id = \
wbase.make_reagent_targeted_gene_id(
gene_id, i)
geno.addReagentTargetedGene(
i, gene_id, targeted_gene_id)
assoc = G2PAssoc(g, self.name, targeted_gene_id,
phenotypeid)
else:
assoc = G2PAssoc(g, self.name, i, phenotypeid)
for r in refs:
r = r.strip()
if r != '':
prefix = re.split(r':', r)[0]
r = re.sub(prefix,
self.clean_db_prefix(prefix), r)
r = re.sub(r'MGI\:MGI\:', 'MGI:', r)
assoc.add_source(r)
# experimental phenotypic evidence
assoc.add_evidence("ECO:0000059")
assoc.add_association_to_graph()
# TODO should the G2PAssoc be
# the evidence for the GO assoc?
if not self.testMode and \
limit is not None and line_counter > limit:
break
return