"""Test partial correlations"""
from unittest import mock
import pytest
from gn3.computations.partial_correlations import partial_correlations_with_target_db
@pytest.mark.integration_test
@pytest.mark.parametrize(
"post_data", (
None, {}, {
"primary_trait": None,
"control_traits": None,
"method": None,
"target_db": None
}, {
"primary_trait": None,
"control_traits": None,
"method": None,
"target_db": "a_db"
}, {
"primary_trait": None,
"control_traits": None,
"method": "a_method",
"target_db": None
}, {
"primary_trait": None,
"control_traits": None,
"method": "a_method",
"target_db": "a_db"
}, {
"primary_trait": None,
"control_traits": ["a_trait", "another"],
"method": None,
"target_db": None
}, {
"primary_trait": None,
"control_traits": ["a_trait", "another"],
"method": None,
"target_db": "a_db"
}, {
"primary_trait": None,
"control_traits": ["a_trait", "another"],
"method": "a_method",
"target_db": None
}, {
"primary_trait": None,
"control_traits": ["a_trait", "another"],
"method": "a_method",
"target_db": "a_db"
}, {
"primary_trait": "a_trait",
"control_traits": None,
"method": None,
"target_db": None
}, {
"primary_trait": "a_trait",
"control_traits": None,
"method": None,
"target_db": "a_db"
}, {
"primary_trait": "a_trait",
"control_traits": None,
"method": "a_method",
"target_db": None
}, {
"primary_trait": "a_trait",
"control_traits": None,
"method": "a_method",
"target_db": "a_db"
}, {
"primary_trait": "a_trait",
"control_traits": ["a_trait", "another"],
"method": None,
"target_db": None
}, {
"primary_trait": "a_trait",
"control_traits": ["a_trait", "another"],
"method": None,
"target_db": "a_db"
}, {
"primary_trait": "a_trait",
"control_traits": ["a_trait", "another"],
"method": "a_method",
"target_db": None
}))
def test_partial_correlation_api_with_missing_request_data(client, post_data):
"""
Test /api/correlations/partial endpoint with various expected request data
missing.
"""
response = client.post("/api/correlation/partial", json=post_data)
assert (
response.status_code == 400 and response.is_json and
response.json.get("status") == "error")
@pytest.mark.integration_test
@pytest.mark.slow
@pytest.mark.parametrize(
"post_data",
({# ProbeSet
"primary_trait": {"dataset": "a_dataset", "trait_name": "a_name"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}, {# Publish
"primary_trait": {
"dataset": "a_Publish_dataset", "trait_name": "a_name"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}, {# Geno
"primary_trait": {"dataset": "a_Geno_dataset", "trait_name": "a_name"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}, {# Temp
"primary_trait": {"dataset": "a_Temp_dataset", "trait_name": "a_name"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}))
def test_partial_correlation_api_with_non_existent_primary_traits(
client, post_data, mocker):
"""
Check that the system responds appropriately in the case where the user
makes a request with a non-existent primary trait.
"""
mocker.patch("gn3.api.correlation.redis.Redis", mock.MagicMock())
response = client.post("/api/correlation/partial", json=post_data)
assert (
response.status_code == 200 and response.is_json and
response.json.get("status") == "success")
@pytest.mark.integration_test
@pytest.mark.slow
@pytest.mark.parametrize(
"post_data",
({# ProbeSet
"primary_trait": {
"dataset": "UCLA_BXDBXH_CARTILAGE_V2",
"trait_name": "ILM103710672"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}, {# Publish
"primary_trait": {"dataset": "BXDPublish", "trait_name": "BXD_12557"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}, {# Geno
"primary_trait": {"dataset": "AKXDGeno", "trait_name": "D4Mit16"},
"control_traits": [
{"dataset": "a_dataset", "trait_name": "a_name"},
{"dataset": "a_dataset2", "trait_name": "a_name2"}],
"method": "a_method",
"target_db": "a_db"
}
# Temp -- the data in the database for these is ephemeral, making it
# difficult to test for this
))
def test_partial_correlation_api_with_non_existent_control_traits(client, post_data, mocker):
"""
Check that the system responds appropriately in the case where the user
makes a request with a non-existent control traits.
The code repetition here is on purpose - valuing clarity over succinctness.
"""
mocker.patch("gn3.api.correlation.redis.Redis", mock.MagicMock())
response = client.post("/api/correlation/partial", json=post_data)
assert (
response.status_code == 200 and response.is_json and
response.json.get("status") == "success")
@pytest.mark.integration_test
@pytest.mark.slow
@pytest.mark.parametrize(
"primary,controls,method,target", (
(# Probeset
"UCLA_BXDBXH_CARTILAGE_V2::ILM103710672", (
"UCLA_BXDBXH_CARTILAGE_V2::nonExisting01",
"UCLA_BXDBXH_CARTILAGE_V2::nonExisting02",
"UCLA_BXDBXH_CARTILAGE_V2::ILM380019"),
"Genetic Correlation, Pearson's r", "BXDPublish"),
(# Publish
"BXDPublish::17937", (
"BXDPublish::17940",
"BXDPublish::nonExisting03"),
"Genetic Correlation, Spearman's rho", "BXDPublish"),
(# Geno
"AKXDGeno::D4Mit16", (
"AKXDGeno::D1Mit170",
"AKXDGeno::nonExisting04",
"AKXDGeno::D1Mit135",
"AKXDGeno::nonExisting05",
"AKXDGeno::nonExisting06"),
"SGO Literature Correlation", "BXDPublish")
)
# Temp -- the data in the database for these is ephemeral, making it
# difficult to test for these without a temp database with the temp
# traits data set to something we are in control of
)
def test_part_corr_api_with_mix_of_existing_and_non_existing_control_traits(
db_conn, primary, controls, method, target):
"""
Check that calling the function with a mix of existing and missing control
traits raises an warning.
"""
criteria = 10
with pytest.warns(UserWarning):
partial_correlations_with_target_db(
db_conn, primary, controls, method, criteria, target)
