def process_gene_interaction(self, limit):
"""
The gene interaction file includes identified interactions,
that are between two or more gene (products).
In the case of interactions with >2 genes, this requires creating
groups of genes that are involved in the interaction.
From the wormbase help list: In the example WBInteraction000007779
it would likely be misleading to suggest that lin-12 interacts with
(suppresses in this case) smo-1 ALONE or that lin-12 suppresses let-60
ALONE; the observation in the paper; see Table V in paper PMID:15990876
was that a lin-12 allele (heterozygous lin-12(n941/+)) could suppress
the "multivulva" phenotype induced synthetically by simultaneous
perturbation of BOTH smo-1 (by RNAi) AND let-60 (by the n2021 allele).
So this is necessarily a three-gene interaction.
Therefore, we can create groups of genes based on their "status" of
Effector | Effected.
Status: IN PROGRESS
:param limit:
:return:
"""
raw = '/'.join((self.rawdir, self.files['gene_interaction']['file']))
if self.testMode:
g = self.testgraph
else:
g = self.graph
gu = GraphUtils(curie_map.get())
logger.info("Processing gene interaction associations")
line_counter = 0
with gzip.open(raw, 'rb') as csvfile:
filereader = csv.reader(
io.TextIOWrapper(csvfile, newline=""), delimiter='\t',
quotechar="'")
for row in filereader:
line_counter += 1
if re.match(r'#', ''.join(row)):
continue
(interaction_num, interaction_type, interaction_subtype,
summary, citation) = row[0:5]
print(row)
interaction_id = 'WormBase:'+interaction_num
# TODO deal with subtypes
interaction_type_id = None
if interaction_type == 'Genetic':
interaction_type_id = \
InteractionAssoc.interaction_object_properties[
'genetically_interacts_with']
elif interaction_type == 'Physical':
interaction_type_id = \
InteractionAssoc.interaction_object_properties[
'molecularly_interacts_with']
elif interaction_type == 'Regulatory':
interaction_type_id = \
InteractionAssoc.interaction_object_properties[
'regulates']
else:
logger.info(
"An interaction type I don't understand %s",
interaction_type)
num_interactors = (len(row) - 5) / 3
if num_interactors != 2:
logger.info(
"Skipping interactions with !=2 participants:\n %s",
str(row))
continue
gene_a_id = 'WormBase:'+row[5]
gene_b_id = 'WormBase:'+row[8]
if self.testMode \
and gene_a_id not in self.test_ids['gene'] \
and gene_b_id not in self.test_ids['gene']:
continue
assoc = InteractionAssoc(
self.name, gene_a_id, gene_b_id, interaction_type_id)
assoc.set_association_id(interaction_id)
assoc.add_association_to_graph(g)
assoc_id = assoc.get_association_id()
# citation is not a pmid or WBref - get this some other way
gu.addDescription(g, assoc_id, summary)
if not self.testMode \
and limit is not None and line_counter > limit:
break
return
def _get_interactions(self, limit):
logger.info("getting interactions")
line_counter = 0
f = '/'.join((self.rawdir, self.files['interactions']['file']))
myzip = ZipFile(f, 'r')
# assume that the first entry is the item
fname = myzip.namelist()[0]
matchcounter = 0
with myzip.open(fname, 'r') as csvfile:
for line in csvfile:
# skip comment lines
if re.match(r'^#', line.decode()):
logger.debug("Skipping header line")
continue
line_counter += 1
line = line.decode().strip()
# print(line)
(interactor_a, interactor_b, alt_ids_a, alt_ids_b, aliases_a,
aliases_b, detection_method, pub_author, pub_id, taxid_a,
taxid_b, interaction_type, source_db, interaction_id,
confidence_val) = line.split('\t')
# get the actual gene ids,
# typically formated like: gene/locuslink:351|BIOGRID:106848
gene_a_num = re.search(
r'locuslink\:(\d+)\|?', interactor_a).groups()[0]
gene_b_num = re.search(
r'locuslink\:(\d+)\|?', interactor_b).groups()[0]
if self.testMode:
g = self.testgraph
# skip any genes that don't match our test set
if (int(gene_a_num) not in self.test_ids) or\
(int(gene_b_num) not in self.test_ids):
continue
else:
g = self.graph
# when not in test mode, filter by taxon
if int(re.sub(r'taxid:', '', taxid_a.rstrip())) not in\
self.tax_ids or\
int(re.sub(
r'taxid:', '', taxid_b.rstrip())) not in\
self.tax_ids:
continue
else:
matchcounter += 1
gene_a = 'NCBIGene:'+gene_a_num
gene_b = 'NCBIGene:'+gene_b_num
# get the interaction type
# psi-mi:"MI:0407"(direct interaction)
int_type = re.search(r'MI:\d+', interaction_type).group()
rel = self._map_MI_to_RO(int_type)
# scrub pubmed-->PMID prefix
pub_id = re.sub(r'pubmed', 'PMID', pub_id)
# remove bogus whitespace
pub_id = pub_id.strip()
# get the method, and convert to evidence code
det_code = re.search(r'MI:\d+', detection_method).group()
evidence = self._map_MI_to_ECO(det_code)
# note that the interaction_id is some kind of internal biogrid
# identifier that does not map to a public URI.
# we will construct a monarch identifier from this
assoc = InteractionAssoc(self.name, gene_a, gene_b, rel)
assoc.add_evidence(evidence)
assoc.add_source(pub_id)
assoc.add_association_to_graph(g)
assoc.load_all_properties(g)
if not self.testMode and (
limit is not None and line_counter > limit):
break
myzip.close()
return
def process_gene_interaction(self, limit):
"""
The gene interaction file includes identified interactions,
that are between two or more gene (products).
In the case of interactions with >2 genes, this requires creating
groups of genes that are involved in the interaction.
From the wormbase help list: In the example WBInteraction000007779
it would likely be misleading to suggest that lin-12 interacts with
(suppresses in this case) smo-1 ALONE or that lin-12 suppresses let-60
ALONE; the observation in the paper; see Table V in paper PMID:15990876
was that a lin-12 allele (heterozygous lin-12(n941/+)) could suppress
the "multivulva" phenotype induced synthetically by simultaneous
perturbation of BOTH smo-1 (by RNAi) AND let-60 (by the n2021 allele).
So this is necessarily a three-gene interaction.
Therefore, we can create groups of genes based on their "status" of
Effector | Effected.
Status: IN PROGRESS
:param limit:
:return:
"""
raw = '/'.join((self.rawdir, self.files['gene_interaction']['file']))
if self.testMode:
g = self.testgraph
else:
g = self.graph
model = Model(g)
logger.info("Processing gene interaction associations")
line_counter = 0
with gzip.open(raw, 'rb') as csvfile:
filereader = csv.reader(io.TextIOWrapper(csvfile, newline=""),
delimiter='\t',
quotechar="'")
for row in filereader:
line_counter += 1
if re.match(r'#', ''.join(row)):
continue
(interaction_num, interaction_type, interaction_subtype,
summary, citation) = row[0:5]
# print(row)
interaction_id = 'WormBase:' + interaction_num
# TODO deal with subtypes
interaction_type_id = None
if interaction_type == 'Genetic':
interaction_type_id = \
InteractionAssoc.interaction_object_properties[
'genetically_interacts_with']
elif interaction_type == 'Physical':
interaction_type_id = \
InteractionAssoc.interaction_object_properties[
'molecularly_interacts_with']
elif interaction_type == 'Regulatory':
interaction_type_id = \
InteractionAssoc.interaction_object_properties[
'regulates']
else:
logger.info("An interaction type I don't understand %s",
interaction_type)
num_interactors = (len(row) - 5) / 3
if num_interactors != 2:
logger.info(
"Skipping interactions with !=2 participants:\n %s",
str(row))
continue
gene_a_id = 'WormBase:' + row[5]
gene_b_id = 'WormBase:' + row[8]
if self.testMode \
and gene_a_id not in self.test_ids['gene'] \
and gene_b_id not in self.test_ids['gene']:
continue
assoc = InteractionAssoc(g, self.name, gene_a_id, gene_b_id,
interaction_type_id)
assoc.set_association_id(interaction_id)
assoc.add_association_to_graph()
assoc_id = assoc.get_association_id()
# citation is not a pmid or WBref - get this some other way
model.addDescription(assoc_id, summary)
if not self.testMode \
and limit is not None and line_counter > limit:
break
return
def _get_interactions(self, limit):
LOG.info("getting interactions")
line_counter = 0
f = '/'.join((self.rawdir, self.files['interactions']['file']))
myzip = ZipFile(f, 'r')
# assume that the first entry is the item
fname = myzip.namelist()[0]
matchcounter = 0
with myzip.open(fname, 'r') as csvfile:
for line in csvfile:
# skip comment lines
if re.match(r'^#', line.decode()):
LOG.debug("Skipping header line")
continue
line_counter += 1
line = line.decode().strip()
# print(line)
(interactor_a, interactor_b, alt_ids_a, alt_ids_b, aliases_a,
aliases_b, detection_method, pub_author, pub_id, taxid_a,
taxid_b, interaction_type, source_db, interaction_id,
confidence_val) = line.split('\t')
taxid_a = taxid_a.rstrip()
taxid_b = taxid_b.rstrip()
# get the actual gene ids,
# typically formated like: gene/locuslink:351|BIOGRID:106848
gene_a_num = re.search(r'locuslink\:(\d+)\|?',
interactor_a).groups()[0]
gene_b_num = re.search(r'locuslink\:(\d+)\|?',
interactor_b).groups()[0]
if self.test_mode:
graph = self.testgraph
# skip any genes that don't match our test set
if (int(gene_a_num) not in self.test_ids) or\
(int(gene_b_num) not in self.test_ids):
continue
else:
graph = self.graph
# when not in test mode, filter by taxon
if taxid_a.split(':')[-1] not in self.tax_ids or \
taxid_b.split(':')[-1] not in self.tax_ids:
continue
else:
matchcounter += 1
gene_a = 'NCBIGene:' + gene_a_num
gene_b = 'NCBIGene:' + gene_b_num
# get the interaction type
# psi-mi:"MI:0407"(direct interaction)
int_type = re.search(r'MI:\d+', interaction_type).group()
rel = self.resolve(int_type, False)
if rel == int_type:
rel = self.globaltt['interacts with']
# scrub pubmed-->PMID prefix
pub_id = re.sub(r'pubmed', 'PMID', pub_id)
# remove bogus whitespace
pub_id = pub_id.strip()
# get the method, and convert to evidence code
det_code = re.search(r'MI:\d+', detection_method).group()
evidence = self.resolve(det_code, False)
if evidence == det_code:
evidence = self.globaltt["experimental evidence"]
# note that the interaction_id is some kind of internal biogrid
# identifier that does not map to a public URI.
# we will construct a monarch identifier from this
assoc = InteractionAssoc(graph, self.name, gene_a, gene_b, rel)
assoc.add_evidence(evidence)
assoc.add_source(pub_id)
assoc.add_association_to_graph()
if not self.test_mode and (limit is not None
and line_counter > limit):
break
myzip.close()
return