def get_gene(self, gene_name):
"""
Get a gene from the cache or attempt to disambiguate or add a new
record.
"""
if not gene_name:
return
gene_pk = self.gene_cache.get(gene_name, None)
if gene_pk:
return Gene(pk=gene_pk)
# Attempt to disambiguate, only if this is the only synonym may it be
# associated.
potential_genes = list(Gene.objects.filter(
synonyms__label__iexact=gene_name).distinct())
if len(potential_genes) == 1:
self.gene_cache[gene_name] = potential_genes[0].pk
return potential_genes[0]
# Only if there are no matches should we create a new record,
# otherwise the synonym war will continue
if len(potential_genes) == 0:
gene = Gene(chr=self.get_chromosome(self.chrom), symbol=gene_name)
gene.save()
self.gene_cache[gene_name] = gene.pk
return gene
def load_families(self, cursor):
cursor.execute('''
SELECT tag, description, gene.id
FROM raw.hgnc_families
LEFT OUTER JOIN gene ON (hgnc_families.hgnc_id = gene.hgnc_id)
ORDER BY tag
''')
keys = ['tag', 'description', 'gene_id']
families = {f.tag: f for f in GeneFamily.objects.all()}
while True:
rows = cursor.fetchmany(100)
if not rows:
break
for row in rows:
record = dict(zip(keys, row))
if record['tag'] not in families:
family = GeneFamily(tag=record['tag'],
description=record['description'])
family.save()
families[family.tag] = family
if record['gene_id']:
gene = Gene(pk=record['gene_id'])
family = families[record['tag']]
gene._state.db = family._state.db
gene.families.add(family)
def setUp(self):
import string
chr1 = Chromosome(value='1', label='1')
chr1.save()
genes = {}
for char in string.lowercase:
g = Gene(chr=chr1, symbol=char)
g.save()
genes[char] = g
geneset = GeneSet(name='test')
geneset.save()
geneset.bulk([genes[c] for c in 'someday'])
self.genes = genes
self.geneset = geneset
def setUp(self):
import string
chr1 = Chromosome(value='1', label='1')
chr1.save()
genes = {}
for char in string.lowercase:
g = Gene(chr=chr1, symbol=char)
g.save()
genes[char] = g
geneset = GeneSet(name='test')
geneset.save()
geneset.bulk([genes[c] for c in 'someday'])
self.genes = genes
self.geneset = geneset
def _get_or_create_gene(self, record):
target = {}
# Link HGNC id
target['hgnc_id'] = int(record['hgnc_id'])
# Parse and map chromosome
if record['chromosome'] == 'mitochondria':
target['chr_id'] = self.chromosomes['MT']
elif ' and ' in record['chromosome']:
target['chr_id'] = self.chromosomes['XY']
else:
match = self.chrom_re.match(record['chromosome'])
if not match:
log.warning('unable to match gene chromosome from HGNC',
extra={
'hgnc_id': record['hgnc_id'],
'raw_chr': record['chromosome'],
})
return None, False
target['chr_id'] = self.chromosomes[match.groups()[0]]
target['symbol'] = record['approved_symbol'].encode('utf8')
target['name'] = record['approved_name'].encode('utf8')
# If the gene already exists by hgnc_id, fetch it. Next check
# by symbol (in case of a new approved gene) and set the
# hgnc_id. Fallback to creating a new gene
try:
return Gene.objects.get(hgnc_id=target['hgnc_id']), False
except Gene.DoesNotExist:
try:
gene = Gene.objects.get(symbol=target['symbol'])
gene.hgnc_id = target['hgnc_id']
except Gene.DoesNotExist:
gene = Gene(**target)
gene.save()
return gene, True
def get_gene(self, gene_name):
"Get a gene from the cache or attempt to disambiguate or add a new record."
if not gene_name:
return
gene_pk = self.gene_cache.get(gene_name, None)
if gene_pk:
return Gene(pk=gene_pk)
# Attempt to disambiguate, only if this is the only synonym may it be
# associated
potential_genes = list(Gene.objects.filter(synonyms__label__iexact=gene_name).distinct())
if len(potential_genes) == 1:
self.gene_cache[gene_name] = potential_genes[0].pk
return potential_genes[0]
# Only if there are no matches should we create a new record,
# otherwise the synonym war will continue
if len(potential_genes) == 0:
gene = Gene(chr=self.get_chromosome(self.chrom), symbol=gene_name)
gene.save()
self.gene_cache[gene_name] = gene.pk
return gene
def _get_or_create_gene(self, record):
target = {}
# Link HGNC id
target["hgnc_id"] = int(record["hgnc_id"])
# Parse and map chromosome
if record["chromosome"] == "mitochondria":
target["chr_id"] = self.chromosomes["MT"]
elif " and " in record["chromosome"]:
target["chr_id"] = self.chromosomes["XY"]
else:
match = self.chrom_re.match(record["chromosome"])
if not match:
log.warning(
"unable to match gene chromosome from HGNC",
extra={"hgnc_id": record["hgnc_id"], "raw_chr": record["chromosome"]},
)
return None, False
target["chr_id"] = self.chromosomes[match.groups()[0]]
target["symbol"] = record["approved_symbol"].encode("utf8")
target["name"] = record["approved_name"].encode("utf8")
# If the gene already exists by hgnc_id, fetch it. Next check
# by symbol (in case of a new approved gene) and set the
# hgnc_id. Fallback to creating a new gene
try:
return Gene.objects.get(hgnc_id=target["hgnc_id"]), False
except Gene.DoesNotExist:
try:
gene = Gene.objects.get(symbol=target["symbol"])
gene.hgnc_id = target["hgnc_id"]
except Gene.DoesNotExist:
gene = Gene(**target)
gene.save()
return gene, True
def _get_or_create_gene(self, record):
target = {}
# Link HGNC id
target['hgnc_id'] = int(record['hgnc_id'])
# Parse and map chromosome
if record['chromosome'] == 'mitochondria':
target['chr_id'] = self.chromosomes['MT']
elif ' and ' in record['chromosome']:
target['chr_id'] = self.chromosomes['XY']
else:
match = self.chrom_re.match(record['chromosome'])
if not match:
log.warning('unable to match gene chromosome from HGNC', extra={
'hgnc_id': record['hgnc_id'],
'raw_chr': record['chromosome'],
})
return None, False
target['chr_id'] = self.chromosomes[match.groups()[0]]
target['symbol'] = record['approved_symbol'].encode('utf8')
target['name'] = record['approved_name'].encode('utf8')
# If the gene already exists by hgnc_id, fetch it. Next check
# by symbol (in case of a new approved gene) and set the
# hgnc_id. Fallback to creating a new gene
try:
return Gene.objects.get(hgnc_id=target['hgnc_id']), False
except Gene.DoesNotExist:
try:
gene = Gene.objects.get(symbol=target['symbol'])
gene.hgnc_id = target['hgnc_id']
except Gene.DoesNotExist:
gene = Gene(**target)
gene.save()
return gene, True
def get_transcript(self, gene_pk, refseq_id):
"Get a transcript from the cache or add a new record."
if not refseq_id:
return
transcript_pk = self.transcripts.get(refseq_id)
if transcript_pk:
return transcript_pk
gene = Gene(pk=gene_pk)
transcript = Transcript(refseq_id=refseq_id, gene=gene)
try:
transcript.save()
except IntegrityError:
transcript = Transcript.objects.get(refseq_id=refseq_id, gene=gene)
self.transcripts[refseq_id] = transcript.pk
return transcript.pk
def load_hgmd_snp(self, cursor, using=None):
cursor.execute('''
select distinct
cl.acc_num as hgmd_id,
c.value as chr,
c.id as chr_id,
v.id as variant_id,
trim(both from cl.disease) as phenotype,
ph.id as phenotype_id,
cl.gene as gene,
g.id as gene_id,
cl.pmid as pubmed,
pm.pmid as pubmed_id
from (
select m.acc_num, m.disease, m.gene, m.pmid, c.chromosome, c."coordSTART" as pos
from raw.hgmd_mutation m inner join raw.hgmd_hg19_coords c on (m.acc_num = c.acc_num)
) cl
left outer join chromosome c on (cl.chromosome = c.value)
left outer join variant v on (c.id = v.chr_id and cl.pos = v.pos)
left outer join variant_type vt on (v.type_id = vt.id)
left outer join phenotype ph on (lower(trim(both from regexp_replace(cl.disease, '\s*\?$', ''))) = lower(ph.term))
left outer join pubmed pm on (cl.pmid::varchar = pm.pmid::varchar)
left outer join gene g on (cl.gene::text = g.symbol::text)
left outer join variant_phenotype vp on (vp.variant_id = v.id)
where vt.value = 'SNP'
and cl.disease not like '%%?'
and v.id is not null and g.id is not null
order by c.id
''')
keys = ['hgmd_id', 'chr', 'chr_id', 'variant_id',
'phenotype', 'phenotype_id', 'gene', 'gene_id',
'pubmed', 'pubmed_id']
count = 0
new_pubmed_map = {}
new_phenotype_map = {}
chrs = dict(Chromosome.objects.values_list('value', 'id'))
while True:
rows = cursor.fetchmany(100)
if not rows:
break
for row in rows:
record = dict(zip(keys, row))
# Get or create a pubmed record
if record['pubmed_id']:
pubmed = PubMed(pmid=record['pubmed_id'])
pubmed._state.db = using
# Some records have a bogus PMID. Only proces the valid ones.
elif type(record['pubmed']) is int or record['pubmed'].isdigit():
pmid = int(record['pubmed'])
if pmid in new_pubmed_map:
pubmed = new_pubmed_map[pmid]
else:
pubmed = PubMed(pmid=pmid)
pubmed.save()
new_pubmed_map[pmid] = pubmed
else:
pubmed = None
# Get or create a the phenotype, associate the HGMD id with
if record['phenotype_id']:
phenotype = Phenotype(pk=record['phenotype_id'])
phenotype._state.db = using
else:
term = record['phenotype']
# Check newly added objects
if term in new_phenotype_map:
phenotype = new_phenotype_map[term]
else:
phenotype = Phenotype(term=record['phenotype'])
phenotype.save()
new_phenotype_map[term] = phenotype
_chr = Chromosome(pk=chrs[record['chr']])
_chr._state.db = using
if record['gene_id']:
gene = Gene(pk=record['gene_id'])
gene._state.db = using
try:
gp = GenePhenotype.objects.get(gene=gene, phenotype=phenotype)
except GenePhenotype.DoesNotExist:
gp = GenePhenotype(gene=gene, phenotype=phenotype)
gp.hgmd_id = record['hgmd_id']
gp.save()
else:
gene = None
if record['variant_id']:
variant = Variant(pk=record['variant_id'])
variant._state.db = using
try:
vp = VariantPhenotype.objects.get(variant=variant, phenotype=phenotype)
except VariantPhenotype.DoesNotExist:
vp = VariantPhenotype(variant=variant, phenotype=phenotype)
vp.hgmd_id = record['hgmd_id']
vp.save()
else:
variant = None
if pubmed:
phenotype.articles.add(pubmed)
if variant:
variant.articles.add(pubmed)
if gene:
gene.articles.add(pubmed)
count += 1
sys.stdout.write('{0}\r'.format(count))
sys.stdout.flush()
return count
