def get_gene_info(ensembl_ids=None, hgnc_symbols=None):
"""Return the genes info based on the transcripts found
Args:
ensembl_ids (Optional[list]): list of Ensembl gene ids
hgnc_symbols (Optional[list]): list of HGNC gene symbols
Returns:
iterable: an iterable with `Gene` objects
"""
uniq_ensembl_ids = set(ensembl_id for ensembl_id in (ensembl_ids or []))
uniq_hgnc_symbols = set(hgnc_symbol for hgnc_symbol in (hgnc_symbols or []))
ensembl = itertools.chain(*[query_gene(ensembl_id=ensembl_id)
for ensembl_id in uniq_ensembl_ids])
hgnc = itertools.chain(*[query_gene(hgnc_symbol=hgnc_symbol)
for hgnc_symbol in uniq_hgnc_symbols])
gene_data = list(itertools.chain.from_iterable([ensembl, hgnc]))
genes = [Gene(symbol=gene['hgnc_symbol'],
hgnc_id=gene['hgnc_id'],
ensembl_id=gene['ensembl_id'],
description=gene['description'],
chrom=gene['chrom'],
start=gene['start'],
stop=gene['stop'],
location=get_cytoband_coord(gene['chrom'], gene['start']),
hi_score=gene['hi_score'],
constraint_score=gene['constraint_score'],
omim_number=get_omim_number(gene['hgnc_symbol']))
for gene in gene_data]
return genes
def get_gene_info(ensembl_ids=None, hgnc_symbols=None):
"""Return the genes info based on the transcripts found
Args:
ensembl_ids (Optional[list]): list of Ensembl gene ids
hgnc_symbols (Optional[list]): list of HGNC gene symbols
Returns:
iterable: an iterable with `Gene` objects
"""
uniq_ensembl_ids = set(ensembl_id for ensembl_id in (ensembl_ids or []))
uniq_hgnc_symbols = set(hgnc_symbol for hgnc_symbol in (hgnc_symbols or []))
genes = []
gene_data = []
if uniq_ensembl_ids:
for ensembl_id in uniq_ensembl_ids:
for res in query_gene(ensembl_id=ensembl_id):
gene_data.append(res)
elif uniq_hgnc_symbols:
for hgnc_symbol in uniq_hgnc_symbols:
query_res = query_gene(hgnc_symbol=hgnc_symbol)
if query_res:
for res in query_res:
gene_data.append(res)
else:
# If no result we add just the symbol
gene_data.append({
'hgnc_symbol': hgnc_symbol,
'hgnc_id': None,
'ensembl_id': None,
'description': None,
'chrom': 'unknown',
'start': 0,
'stop': 0,
'hi_score': None,
'constraint_score': None,
})
for gene in gene_data:
genes.append(Gene(
symbol=gene ['hgnc_symbol'],
hgnc_id=gene['hgnc_id'],
ensembl_id=gene['ensembl_id'],
description=gene['description'],
chrom=gene['chrom'],
start=gene['start'],
stop=gene['stop'],
location=get_cytoband_coord(gene['chrom'], gene['start']),
hi_score=gene['hi_score'],
constraint_score=gene['constraint_score'],
omim_number=get_omim_number(gene['hgnc_symbol'])
))
return genes
def get_gene_info(ensembl_ids=None, hgnc_symbols=None):
"""Return the genes info based on the transcripts found
Args:
ensembl_ids (Optional[list]): list of Ensembl gene ids
hgnc_symbols (Optional[list]): list of HGNC gene symbols
Returns:
iterable: an iterable with `Gene` objects
"""
uniq_ensembl_ids = set(ensembl_id for ensembl_id in (ensembl_ids or []))
uniq_hgnc_symbols = set(hgnc_symbol
for hgnc_symbol in (hgnc_symbols or []))
genes = []
gene_data = []
if uniq_ensembl_ids:
for ensembl_id in uniq_ensembl_ids:
for res in query_gene(ensembl_id=ensembl_id):
gene_data.append(res)
elif uniq_hgnc_symbols:
for hgnc_symbol in uniq_hgnc_symbols:
query_res = query_gene(hgnc_symbol=hgnc_symbol)
if query_res:
for res in query_res:
gene_data.append(res)
else:
# If no result we add just the symbol
gene_data.append({
'hgnc_symbol': hgnc_symbol,
'hgnc_id': None,
'ensembl_id': None,
'description': None,
'chrom': 'unknown',
'start': 0,
'stop': 0,
'hi_score': None,
'constraint_score': None,
})
for gene in gene_data:
genes.append(
Gene(symbol=gene['hgnc_symbol'],
hgnc_id=gene['hgnc_id'],
ensembl_id=gene['ensembl_id'],
description=gene['description'],
chrom=gene['chrom'],
start=gene['start'],
stop=gene['stop'],
location=get_cytoband_coord(gene['chrom'], gene['start']),
hi_score=gene['hi_score'],
constraint_score=gene['constraint_score'],
omim_number=get_omim_number(gene['hgnc_symbol'])))
return genes
def get_gene_info(ensembl_ids=None, hgnc_symbols=None):
"""Return the genes info based on the transcripts found
Args:
transcript(Transcript): A dictionary with transcript info
Returns:
genes (iterable): An iterable with Genes
"""
if ensembl_ids:
ensembl_ids = set([ens_id for ens_id in ensembl_ids])
elif hgnc_symbols:
hgnc_symbols = set([symbol for symbol in hgnc_symbols])
genes = []
if ensembl_ids:
for ensembl_id in ensembl_ids:
if ensembl_id:
for gene in query_gene(ensembl_id=ensembl_id):
genes.append(Gene(
symbol=gene['hgnc_symbol'],
hgnc_id=gene['hgnc_id'],
ensembl_id=gene['ensembl_id'],
description=gene['description'],
chrom=gene['chrom'],
start=gene['start'],
stop=gene['stop'],
location=get_cytoband_coord(gene['chrom'], gene['start']),
hi_score=gene['hi_score'],
constraint_score=gene['constraint_score'],
omim_number=get_omim_number(gene['hgnc_symbol']),
))
elif hgnc_symbols:
for hgnc_symbol in hgnc_symbols:
if hgnc_symbol:
for gene in query_gene(hgnc_symbol=hgnc_symbol):
genes.append(Gene(
symbol=gene['hgnc_symbol'],
hgnc_id=gene['hgnc_id'],
ensembl_id=gene['ensembl_id'],
description=gene['description'],
chrom=gene['chrom'],
start=gene['start'],
stop=gene['stop'],
location=get_cytoband_coord(gene['chrom'], gene['start']),
hi_score=gene['hi_score'],
constraint_score=gene['constraint_score'],
omim_number=get_omim_number(gene['hgnc_symbol']),
))
return genes
def test_query_ensembl_id(database):
ensembl_id = 'ENSG00000070814'
hgnc_symbol = 'TCOF1'
result = query_gene(ensembl_id=ensembl_id, connection=database)
gene = result[0]
assert gene['ensembl_id'] == ensembl_id
assert gene['hgnc_symbol'] == hgnc_symbol
def test_query_ensembl_id(database):
ensembl_id = 'ENSG00000070814'
hgnc_symbol = 'TCOF1'
result = query_gene(ensembl_id=ensembl_id, connection=database)
gene = result[0]
assert gene['ensembl_id'] == ensembl_id
assert gene['hgnc_symbol'] == hgnc_symbol
def test_query_non_existing_term(database):
"""docstring for query_term"""
term = 'ENSG00000000004'
result = query_gene(term, connection=database)
assert result == []
def test_query_wrong_term(database):
"""docstring for query_term"""
term = 'HPO'
with pytest.raises(ValueError):
result = query_gene(term, connection=database)
def test_query_no_term(database):
"""docstring for query_term"""
with pytest.raises(SyntaxError):
result = query_gene(connection=database)
def test_query_non_existing_term(database):
"""docstring for query_term"""
term = 'ENSG00000000004'
result = query_gene(term, connection=database)
assert result == []
def test_query_wrong_term(database):
"""docstring for query_term"""
term = 'HPO'
with pytest.raises(ValueError):
result = query_gene(term, connection=database)
def test_query_no_term(database):
"""docstring for query_term"""
with pytest.raises(SyntaxError):
result = query_gene(connection=database)
