def getEnsemblGene(gene):
"""Find ENSG for this gene.
Look in HGNC aliases for this gene name.
If a page has no ENGG, use the HGNCs
with the hgnc module to look for one."""
searchUrl = baseUrl + 'Search/Keyword?queryString=%s' % (gene, )
response = requests.get(searchUrl)
soup = bs4.BeautifulSoup(response.text)
links = soup.find_all("td", class_="gc-gene-symbol")
ensemblGeneSet = set()
i = 0
for link in links:
for item in link:
name = str(item).split('data-ga-label="')[1].split('"')[0]
href = baseUrl + str(item).split('href="')[1].split('"')[0].split('&')[0]
time.sleep(2)
newHtml = requests.get(href)
hgncAliasSet = getPreviousHGNC(newHtml) | getAlias(newHtml)
ensemblSet = mkEnsemblSet(newHtml) | set([hgnc.fetchEnsemblGeneIdForGeneSymbol(name)])
if name == gene or gene in hgncAliasSet:
ensemblGeneSet |= ensemblSet
if i:
time.sleep(2)
i += 1
return ensemblGeneSet
def testSearchSymbol():
nose.tools.assert_equal(hgnc.fetchEnsemblGeneIdForGeneSymbol('RBM5-AS1'),
'ENSG00000281691')
