def source_from_id(acc):
client = DatasetClient()
api_output = client.search(acc, "id")
if api_output.datasets:
datasets = api_output.datasets
if len(datasets) > 1:
hits = []
for dataset in datasets:
if dataset.accession == acc:
hits.append(dataset)
if len(hits) == 1:
source = hits[0].database
return source
else:
click.echo('INFO')
click.echo(
f'{acc} is not a unique ID, leading to multiple hits.')
sys.exit()
else:
source = datasets[0].database
return source
else:
click.echo('INFO')
click.echo(f'{acc} could not be found in one of the repositories.')
sys.exit()
def test_get_object_from_json(self):
client = DatasetClient()
dataset = client.get_dataset_details("pride", "PXD000210")
# dataset = Dataset.get_object_from_json(object_json)
assert dataset.accession == "PXD000210"
assert dataset.title == "Proteome analysis by charge state-selective separation of peptides: a multidimensional approach"
assert len(dataset.dates) > 0
assert len(dataset.scores) > 0
assert len(dataset.cross_references) > 0
assert len(dataset.keywords) > 0
assert len(dataset.organisms) > 0
assert len(dataset.get_posttranslational_modifications()) == 7
assert len(dataset.get_diseases()) == 0
assert len(dataset.get_tissues()) == 0
dataset_transcriptomics = client.get_dataset_details("arrayexpress-repository", "E-TABM-555")
# dataset_transcriptomics = Dataset.get_object_from_json(object_json)
assert dataset_transcriptomics.accession == "E-TABM-555"
assert dataset_transcriptomics.title == "Transcription profiling of rat to investigate technical and biological variability on the Agilent platform"
assert len(dataset_transcriptomics.dates) > 0
assert len(dataset_transcriptomics.scores) > 0
assert len(dataset_transcriptomics.cross_references) > 0
assert len(dataset_transcriptomics.keywords) > 0
assert len(dataset_transcriptomics.organisms) > 0
assert len(dataset_transcriptomics.get_posttranslational_modifications()) == 0
assert len(dataset_transcriptomics.get_diseases()) == 0
assert len(dataset_transcriptomics.get_tissues()) == 4
def test_most_accessed(self):
client = DatasetClient()
res = client.most_accessed(20)
assert res.status_code == 200
assert res.json()["count"] == 20
try:
res = client.most_accessed(0)
except BadRequest as err:
assert err.status == 500
def test_latest(self):
client = DatasetClient()
res = client.latest(20)
assert res.status_code == 200
assert res.json()["count"] > 0
try:
client.latest(0)
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
def test_most_accessed(self):
client = DatasetClient()
res = client.most_accessed(20)
assert res.count > 0
assert len(res.datasets) > 0
try:
res = client.most_accessed(0)
except BadRequest as err:
assert err.status == 500
def test_latest(self):
client = DatasetClient()
res = client.latest(20)
assert res.count > 0
assert len(res.datasets) > 0
try:
client.latest(0)
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
def file_links(source, acc):
client = DatasetClient()
api_output = client.get_dataset_details(source, acc)
file_links = {}
if api_output.files:
files = api_output.files[0]['files']
else:
click.echo('INFO')
click.echo(
f'The accession {acc} is found in the {source} database, but contains no files.'
)
sys.exit()
return files
def test_get_similar_by_pubmed(self):
client = DatasetClient()
res = client.get_similar_by_pubmed("16585740")
assert res.status_code == 200
try:
res = client.get_similar_by_pubmed("qq9q9")
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
try:
res = client.get_similar_by_pubmed(None)
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_get_similar(self):
client = DatasetClient()
res = client.get_similar("PXD000210", "pride")
assert res.status_code == 200
try:
res = client.get_similar("aaaa", "pride")
except BadRequest as err:
assert err.status == 500
try:
res = client.get_similar(None, "pride")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_get_file_links(self):
client = DatasetClient()
res = client.get_file_links("PXD000210", "pride")
assert res.status_code == 200
assert len(res.json()) > 0
try:
res = client.get_file_links("aaa", "pride")
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
try:
res = client.get_file_links(None, "pride")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_batch(self):
client = DatasetClient()
res = client.batch("PXD000210", 'pride')
assert res.status_code == 200
assert res.json()["datasets"][0]["id"] == "PXD000210"
try:
client.batch("gulugulu11", "momomomomo")
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
try:
client.batch(None, "momomomomo")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_search(self):
client = DatasetClient()
res = client.search("cancer human", "publication_date", "ascending")
assert res.json()["count"] > 0
try:
dataset = client.search("j9j9j9j9@£", "publication_date",
"ascending")
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
try:
dataset = client.search(None, "publication_date", "ascending")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_batch(self):
client = DatasetClient()
res = client.batch("PXD000210", 'pride')
assert len(res.failure) == 0
assert res.datasets[0].accession == "PXD000210"
try:
client.batch("gulugulu11", "momomomomo")
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
try:
client.batch(None, "momomomomo")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_get_dataset_details(self):
"""
check the details for one dataset.
:return:
"""
client = DatasetClient()
dataset = client.get_dataset_details("pride", "PXD000210", False)
assert dataset['accession'] == "PXD000210"
assert len(dataset['description']) == 2227
try:
dataset = client.get_dataset_details("PXD@£", "pride")
except BadRequest as err:
assert err.status == 500
try:
dataset = client.get_dataset_details(None, "pride")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
def test_search(self):
client = DatasetClient()
res = client.search("cancer human", "publication_date", "ascending")
assert len(res.datasets) > 0
assert res.count > 0
res_other_pages = client.search("cancer human", "publication_date", "ascending", 1200, 30, 20)
assert res_other_pages.count > 0
assert len(res_other_pages.datasets) == 30
try:
dataset = client.search("j9j9j9j9@£", "publication_date", "ascending")
except BadRequest as err:
assert err.status == DATA_NOT_FOUND
try:
dataset = client.search(None, "publication_date", "ascending")
except BadRequest as err:
assert err.status == MISSING_PARAMETER
