def search_data(free_text_search, query, result, display, download, file,
fields, sortfields, offset, length):
""" Search data given a query """
free_text_search = True if free_text_search else False
download = None if not download else download
file = None if not file else file
fields = None if not fields else ",".join(fields)
sortfields = None if not sortfields else ",".join(sortfields)
offset = None if not offset else offset
length = None if not length else length
if display in ["fasta", "fastq"]:
results = enasearch.search_all_data(free_text_search=free_text_search,
query=query,
result=result,
display=display,
download=download,
file=file)
else:
results = enasearch.search_data(free_text_search=free_text_search,
query=query,
result=result,
display=display,
download=download,
file=file,
fields=fields,
sortfields=sortfields,
offset=offset,
length=length)
if file is None:
print_display(results, display)
def test_search_data():
"""Test search_data function"""
search_data = enasearch.search_data(
free_text_search=False,
query="tax_tree(7147) AND dataclass=STD",
result="coding_release",
display="fasta",
offset=0,
length=20,
download=None,
file=None,
fields=None,
sortfields=None)
assert len([seq.id for seq in search_data]) > 0
def search_data(free_text_search, query, result, display, download, file,
fields, sortfields, offset, length):
"""Search data given a query.
This function
- Extracts the number of possible results for the query
- Extracts the all the results of the query (by potentially running several times the search function)
The output can be redirected to a file and directly display to the standard
output given the display chosen.
"""
free_text_search = True if free_text_search else False
download = None if not download else download
file = None if not file else file
fields = None if not fields else ",".join(fields)
sortfields = None if not sortfields else ",".join(sortfields)
offset = None if not offset else offset
length = None if not length else length
if display in ["fasta", "fastq"]:
results = enasearch.search_all_data(free_text_search=free_text_search,
query=query,
result=result,
display=display,
download=download,
file=file)
else:
results = enasearch.search_data(free_text_search=free_text_search,
query=query,
result=result,
display=display,
download=download,
file=file,
fields=fields,
sortfields=sortfields,
offset=offset,
length=length)
if file is None:
print_display(results, display)
data = json.loads(out)
return data[0]
def sample2assembly(sample_id):
response = ena_sample(sample_id, result='assembly')
if response.status_code != 200:
return ''
else:
out = response.text
data = json.loads(out)
return data[0]
data = enasearch.search_data(free_text_search=True,
query="CABMLH010000000",
result='wgs_set',
display='xml')
data = enasearch.retrieve_data(ids="CABMLH010000000", display="html")
def get_returnable_fields(result):
url = f"https://www.ebi.ac.uk/ena/portal/api/returnFields?dataPortal=ena&format=json&result={result}"
response = requests.get(url, )
out = response.text
data = json.loads(out)
return data
returnable_fields = get_returnable_fields('wgs_set')
returnable_fields = [_['columnId'] for _ in returnable_fields]