def getSRARuns(orgname, platform):
"""Return a list of run accessions for an organism given a sequencing platform."""
assert platform in ['454', 'Illumina']
term = "%s[Organism] AND transcriptom*" % orgname
search_params = {'db':'sra', 'term':term, 'usehistory':'y'}
root = ElementTree.XML(getURL(ESEARCH_URL, search_params))
result_count = root.findtext("Count")
if int(result_count) == 0:
return []
search_params['WebEnv'] = root.findtext('WebEnv')
search_params['retmax'] = 20
runs = []
sraxmlparser = SRAXMLParser()
for i in range(0, int(result_count), search_params['retmax']):
search_params['retstart'] = str(i)
root = ElementTree.XML(getURL(ESEARCH_URL, search_params))
idlist = []
for idelement in root.findall('IdList/Id'):
idlist.append(idelement.text.strip())
fetch_params = {'db':'sra', 'id':','.join(idlist)}
sraruns = sraxmlparser.parseFromString(getURL(EFETCH_URL, fetch_params))
for srarun in sraruns:
if srarun.platform == platform:
runs.append(srarun)
return runs
def isPaired(srafetchxml, accession): """Check if the run with accession is paired.""" if accession and \ srafetchxml and \ os.path.exists(srafetchxml): sraxmlparser = SRAXMLParser() runs = sraxmlparser.parse(srafetchxml) for run in runs: if run.accession.strip() == accession.strip(): return run.is_paired return False
def getRunsByPlatform(cf):
"""Write the run accessions from a particular platform to a file."""
platform = cf.get_parameter('platform', 'string')
if not platform in ['454', 'Illumina']:
cf.write_error("Unknown sequencing platform %s" % platform)
return constants.GENERIC_ERROR
srafetchxml = cf.get_input('srafetchxml')
srarunlist = cf.get_output('srarunlist')
sraxmlparser = SRAXMLParser()
runs = sraxmlparser.parse(srafetchxml)
writer = csv.writer(open(srarunlist, 'wb'), quoting=csv.QUOTE_NONE)
writer.writerow(['NCBISRARunID'])
num_accessions = 0
for run in runs:
if run.platform == platform:
writer.writerow([run.accession])
num_accessions += 1
cf.write_log("GetRunsByPlatform: wrote %s run accessions" % num_accessions)
return constants.OK
def getExpAccessions(cf):
"""Write the experiment accessions to a file."""
platform = cf.get_parameter('platform')
srafetchxml = cf.get_input('srafetchxml')
sraexplist = cf.get_output('sraexplist')
sraxmlparser = SRAXMLParser()
runs = sraxmlparser.parse(srafetchxml)
writer = csv.writer(open(sraexplist, 'wb'), quoting=csv.QUOTE_NONE)
writer.writerow(['NCBISRAExpID'])
accessions = []
for run in runs:
if platform and \
not run.platform == platform:
continue
elif not run.exp_accession in accessions:
writer.writerow([run.exp_accession])
accessions.append(run.exp_accession)
cf.write_log("GetExpAccessions: wrote %s experiment accessions" % len(accessions))
return constants.OK
def getRunsByExp(cf):
"""Write the runs accessions for a given experiment to a file."""
exp_accession = cf.get_parameter('accession', 'string')
platform = cf.get_parameter('platform', 'string')
srafetchxml = cf.get_input('srafetchxml')
srarunlist = cf.get_output('srarunlist')
sraxmlparser = SRAXMLParser()
runs = sraxmlparser.parse(srafetchxml)
writer = csv.writer(open(srarunlist, 'wb'), quoting=csv.QUOTE_NONE)
writer.writerow(['NCBISRARunID'])
index = 0
for run in runs:
if platform and \
not run.platform == platform:
continue
elif run.exp_accession == exp_accession:
writer.writerow([run.accession])
index += 1
cf.write_log("GetRunsByExp: wrote %s run accessions" % index)
return constants.OK
