def run_sample(args: object):
"""
Runs sample ID through snakemake pipeline
"""
sample: list[dict] = datahandling.get_samples(
sample_ids=[args.sample_id]
) #this syntax is type annotation. That is sample (the annotation target) is of type list[dict]. this is done before variable assignment and can be useful for use with type checkers
component: list[dict] = datahandling.get_components(
component_names=[COMPONENT['name']],
component_versions=[COMPONENT['version']])
if len(component) == 0:
print(
f"component not found in DB, would you like to install it (Y/N)?:")
install: string = input()
if install.upper() == "Y":
datahandling.post_component(COMPONENT)
component: list[dict] = datahandling.get_components(
component_names=[COMPONENT['name']],
component_versions=[COMPONENT['version']])
if len(component) != 1:
print(
f"Error with installation of {COMPONENT['name']} v:{COMPONENT['version']} \n"
)
exit()
else:
print(f"To continue please install the component")
exit()
if len(sample) == 0:
# Invalid sample id
print(f"sample_id not found in DB")
pass
elif len(sample) != 1:
print(f"Error with sample_id")
elif len(component) != 1:
print(f"Error with component_id")
else:
print(
f"snakemake -s /bifrost/{COMPONENT['name']}/pipeline.smk --config sample_id={str(sample[0]['_id'])} component_id={str(component[0]['_id'])}"
)
try:
process: subprocess.Popen = subprocess.Popen(
f"snakemake -s /bifrost/{COMPONENT['name']}/pipeline.smk --config sample_id={str(sample[0]['_id'])} component_id={str(component[0]['_id'])}",
stdout=sys.stdout,
stderr=sys.stderr,
shell=True)
process.communicate()
except:
print(traceback.format_exc())
def test_delete_run():
sample = {"name": "test_sample"}
sample_db = datahandling.post_sample(sample)
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
s_c = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db["_id"],
"name": component_db["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
datahandling.post_sample_component(s_c)
run = {
"name": "test_run",
"samples": [{
"_id": sample_db["_id"]
}],
"components": [{
"_id": component_db["_id"]
}
# Missing name
]
}
run_db = datahandling.post_run(run)
deleted = datahandling.delete_run(run_id=str(run_db["_id"]))
assert deleted == 1
def test_get_sample_using_component_id():
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
sample = {"name": "test_sample", "components": [component_db]}
sample_db = datahandling.post_sample(sample)
sample["_id"] = sample_db["_id"]
sample_received = datahandling.get_samples(
component_ids=[str(component_db["_id"])])[0]
assert sample_received == sample
def install_component():
component: list[dict] = datahandling.get_components(
component_names=[COMPONENT['name']],
component_versions=[COMPONENT['version']])
if len(component) == 1:
print(f"Component has already been installed")
elif len(component) > 1:
print(
f"Component exists multiple times in DB, please contact an admin to fix this in order to proceed"
)
else:
datahandling.post_component(COMPONENT)
component: list[dict] = datahandling.get_components(
component_names=[COMPONENT['name']],
component_versions=[COMPONENT['version']])
if len(component) != 1:
print(
f"Error with installation of {COMPONENT['name']} v:{COMPONENT['version']} \n"
)
exit()
def test_post_sample_component():
sample = {"name": "test_sample"}
sample_db = datahandling.post_sample(sample)
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
s_c = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db["_id"]
},
"results": {},
"summary": {}
}
s_c_db = datahandling.post_sample_component(s_c)
s_c["_id"] = s_c_db["_id"]
# we can't get this timestamp so we won't test it
s_c_db.pop('metadata', None)
s_c.pop('metadata', None)
assert s_c == s_c_db
def test_get_sample_components():
sample = {"name": "test_sample"}
sample_db = datahandling.post_sample(sample)
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
s_c = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db["_id"],
"name": component_db["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
s_c_db = datahandling.post_sample_component(s_c)
# Testing this.
s_c_2 = datahandling.get_sample_components(
sample_ids=[str(sample_db["_id"])], component_names=["assemblatron"])
assert s_c_2[0] == s_c_db
def test_delete_sample():
sample = {"name": "test_sample"}
sample_db = datahandling.post_sample(sample)
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
s_c = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db["_id"],
"name": component_db["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
datahandling.post_sample_component(s_c)
# Testing this
deleted = datahandling.delete_sample(str(sample_db["_id"]))
assert deleted == 1
def test_delete_component():
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
deleted = datahandling.delete_component(str(component_db["_id"]))
assert deleted == 1
def test_post_component():
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
component["_id"] = component_db["_id"]
assert component == component_db
def test_import_run():
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
component_2 = {"name": "whats_my_species"}
component_db_2 = datahandling.post_component(component_2)
sample = {
"name":
"test_sample",
"components": [{
"_id": component_db["_id"]
}, {
"_id": component_db_2["_id"]
}]
}
sample_db = datahandling.post_sample(sample)
s_c = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db["_id"],
"name": component_db["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
datahandling.post_sample_component(s_c)
time.sleep(0.2)
# we need to wait otherwise they share the same setup_date and
# it screws up the order
s_c_2 = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db_2["_id"],
"name": component_db_2["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
datahandling.post_sample_component(s_c_2)
run = {
"name":
"test_run",
"samples": [{
"_id": sample_db["_id"]
}],
"components": [{
"_id": component_db["_id"]
}, {
"_id": component_db_2["_id"]
}
# Missing name
]
}
run_db = datahandling.post_run(run)
run_export = datahandling.get_run_export(names=[str(run_db["name"])])
#drop database
conn = mongomock.MongoClient('mongodb://server.example.com:27017')
conn.drop_database("serumqc_prod")
conn.close()
datahandling.post_run_export(run_export)
get_items = {
"samples": [],
"sample_components": [],
"runs": [],
"components": []
}
run_dict = run_export[str(run_db["name"])]
for s in run_dict["samples"]:
get_s = datahandling.get_samples(sample_ids=[str(s["_id"])])[0]
get_items["samples"].append(get_s)
for s_c in run_dict["sample_components"]:
get_s_c = datahandling.get_sample_components(
sample_component_ids=[str(s_c["_id"])])[0]
get_items["sample_components"].append(get_s_c)
for r in run_dict["runs"]:
get_r = datahandling.get_runs(run_id=str(r["_id"]))[0]
get_items["runs"].append(get_r)
for c in run_dict["components"]:
get_c = datahandling.get_components(component_ids=[str(c["_id"])])[0]
get_items["components"].append(get_c)
assert get_items == run_dict
def test_export_run():
component = {"name": "assemblatron"}
component_db = datahandling.post_component(component)
component_2 = {"name": "whats_my_species"}
component_db_2 = datahandling.post_component(component_2)
sample = {
"name":
"test_sample",
"components": [{
"_id": component_db["_id"]
}, {
"_id": component_db_2["_id"]
}]
}
sample_db = datahandling.post_sample(sample)
s_c = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db["_id"],
"name": component_db["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
s_c_db = datahandling.post_sample_component(s_c)
time.sleep(0.2)
# we need to wait otherwise they share the same setup_date and
# it screws up the order
s_c_2 = {
"sample": {
"_id": sample_db["_id"]
},
"component": {
"_id": component_db_2["_id"],
"name": component_db_2["name"]
},
"results": {},
"summary": {},
"setup_date": datetime.datetime.now()
}
s_c_db_2 = datahandling.post_sample_component(s_c_2)
run = {
"name":
"test_run",
"samples": [{
"_id": sample_db["_id"]
}],
"components": [{
"_id": component_db["_id"]
}, {
"_id": component_db_2["_id"]
}
# Missing name
]
}
run_db = datahandling.post_run(run)
run_export = datahandling.get_run_export(names=[str(run_db["name"])])
run_id = str(run_db["_id"])
run_expected = {
run_db["name"]: {
"components": [component_db_2, component_db],
"samples": [sample_db],
"sample_components": [s_c_db_2, s_c_db],
"runs": [run_db]
}
}
assert run_export == run_expected