def test_retrieve_run_report():
"""Test retrieve_run_report function"""
report = enasearch.retrieve_run_report(accession="SRX017289",
fields=None,
file=None)
assert "small RNAs_wild type" in report
exp_fields = ["run_accession", "fastq_ftp", "fastq_md5", "fastq_bytes"]
report = enasearch.retrieve_run_report(accession="SRX017289",
fields=",".join(exp_fields),
file=None)
assert cmp(report.split("\n")[0].split("\t"), exp_fields)
def get_fastq_urls(accessions: List[str],
fields: List[str] = None) -> Dict[str, Dict]:
"""
Given an ENA (or SRA) Run (SRR*), Experiment (SRX*), Project (PRJ*)
(or Study?) accession, return a list of associated FASTQ download URLs.
:param fields:
:type fields:
:param accessions:
:type accessions:
:return:
:rtype:
"""
if fields is None:
fields = [
'run_accession', 'experiment_accession', 'study_accession',
'sample_accession', 'secondary_sample_accession',
'instrument_platform', 'library_strategy', 'read_count',
'fastq_ftp', 'fastq_md5', 'fastq_bytes'
]
urls_dict = dict()
# raises HTTPError on status_code 500 (eg ENA is temporarily down)
for accession in accessions:
table = enasearch.retrieve_run_report(accession=accession,
fields=','.join(fields))
table = flatten_fastq_table(table)
urls = parse_fastq_table(table, key_by='fastq_ftp')
urls_dict.update(urls)
return urls_dict
def retrieve_run_report(accession, fields, file):
""" Retrieve run report """
fields = None if not fields else ",".join(fields)
file = None if not file else file
report = enasearch.retrieve_run_report(accession=accession,
fields=fields,
file=file)
if file is None:
print_display(report, 'report')
def retrieve_run_report(accession, fields, file):
"""Retrieve run report from ENA.
The output can be redirected to a file and directly display to the standard
output given the display chosen.
"""
fields = None if not fields else ",".join(fields)
file = None if not file else file
report = enasearch.retrieve_run_report(accession=accession,
fields=fields,
file=file)
if file is None:
print_display(report, 'report')
def get_run_table(accessions: List[str],
fields: List[str] = None) -> Dict[str, Dict]:
"""
Given an ENA (or SRA) Run (SRR*), Experiment (SRX*), Project (PRJ*)
(or Study?) accession, return a list of associated Sample-style records
(including FASTQ download URLs) for each run suitable for use by the
frontend.
:param fields:
:type fields:
:param accessions:
:type accessions:
:return:
:rtype:
"""
if fields is None:
fields = [
'run_accession', 'experiment_accession', 'study_accession',
'sample_accession', 'secondary_sample_accession',
'instrument_platform', 'instrument_model', 'library_strategy',
'library_source', 'library_layout', 'library_selection',
'library_name', 'broker_name', 'study_alias', 'experiment_alias',
'sample_alias', 'run_alias', 'read_count', 'base_count',
'fastq_ftp', 'fastq_md5', 'fastq_bytes', 'center_name'
]
runs_dict = dict()
# raises HTTPError on status_code 500 (eg ENA is temporarily down)
for accession in accessions:
table = enasearch.retrieve_run_report(accession=accession,
fields=','.join(fields))
# table = flatten_fastq_table(table)
runs = parse_fastq_table(table, key_by='run_accession')
runs_dict.update(runs)
# We turn the list of FTP urls into a list of dicts like
# [{'R1': 'ftp://bla_1.fastq.gz'}, {'R2': 'ftp://bla_2.fastq.gz'}]
for run, metadata in runs_dict.items():
if metadata.get('fastq_ftp', False):
metadata['fastq_ftp'] = [{
'R%s' % str(n + 1): url
} for n, url in enumerate(metadata['fastq_ftp'])]
return runs_dict