Merge branch 'partial-correlations'

2 files changed, 272 insertions, 0 deletions

diff --git a/tests/unit/db/test_correlation.py b/tests/unit/db/test_correlation.py
new file mode 100644
index 0000000..3f940b2
--- /dev/null
+++ b/tests/unit/db/test_correlation.py
@@ -0,0 +1,96 @@
+"""
+Tests for the gn3.db.correlations module
+"""
+
+from unittest import TestCase
+
+from gn3.db.correlations import (
+    build_query_sgo_lit_corr,
+    build_query_tissue_corr)
+
+class TestCorrelation(TestCase):
+    """Test cases for correlation data fetching functions"""
+    maxDiff = None
+
+    def test_build_query_sgo_lit_corr(self):
+        """
+        Test that the literature correlation query is built correctly.
+        """
+        self.assertEqual(
+            build_query_sgo_lit_corr(
+                "Probeset",
+                "temp_table_xy45i7wd",
+                "T1.value, T2.value, T3.value",
+                (("LEFT JOIN ProbesetData AS T1 "
+                  "ON T1.Id = ProbesetXRef.DataId "
+                  "AND T1.StrainId=%(T1_sample_id)s"),
+                 (
+                     "LEFT JOIN ProbesetData AS T2 "
+                     "ON T2.Id = ProbesetXRef.DataId "
+                     "AND T2.StrainId=%(T2_sample_id)s"),
+                 (
+                     "LEFT JOIN ProbesetData AS T3 "
+                     "ON T3.Id = ProbesetXRef.DataId "
+                     "AND T3.StrainId=%(T3_sample_id)s"))),
+            (("SELECT Probeset.Name, temp_table_xy45i7wd.value, "
+              "T1.value, T2.value, T3.value "
+              "FROM (Probeset, ProbesetXRef, ProbesetFreeze) "
+              "LEFT JOIN temp_table_xy45i7wd ON temp_table_xy45i7wd.GeneId2=ProbeSet.GeneId "
+              "LEFT JOIN ProbesetData AS T1 "
+              "ON T1.Id = ProbesetXRef.DataId "
+              "AND T1.StrainId=%(T1_sample_id)s "
+              "LEFT JOIN ProbesetData AS T2 "
+              "ON T2.Id = ProbesetXRef.DataId "
+              "AND T2.StrainId=%(T2_sample_id)s "
+              "LEFT JOIN ProbesetData AS T3 "
+              "ON T3.Id = ProbesetXRef.DataId "
+              "AND T3.StrainId=%(T3_sample_id)s "
+              "WHERE ProbeSet.GeneId IS NOT NULL "
+              "AND temp_table_xy45i7wd.value IS NOT NULL "
+              "AND ProbesetXRef.ProbesetFreezeId = ProbesetFreeze.Id "
+              "AND ProbesetFreeze.Name = %(db_name)s "
+              "AND Probeset.Id = ProbesetXRef.ProbesetId "
+              "ORDER BY Probeset.Id"),
+             2))
+
+    def test_build_query_tissue_corr(self):
+        """
+        Test that the tissue correlation query is built correctly.
+        """
+        self.assertEqual(
+            build_query_tissue_corr(
+                "Probeset",
+                "temp_table_xy45i7wd",
+                "T1.value, T2.value, T3.value",
+                (("LEFT JOIN ProbesetData AS T1 "
+                  "ON T1.Id = ProbesetXRef.DataId "
+                  "AND T1.StrainId=%(T1_sample_id)s"),
+                 (
+                     "LEFT JOIN ProbesetData AS T2 "
+                     "ON T2.Id = ProbesetXRef.DataId "
+                     "AND T2.StrainId=%(T2_sample_id)s"),
+                 (
+                     "LEFT JOIN ProbesetData AS T3 "
+                     "ON T3.Id = ProbesetXRef.DataId "
+                     "AND T3.StrainId=%(T3_sample_id)s"))),
+            (("SELECT Probeset.Name, temp_table_xy45i7wd.Correlation, "
+              "temp_table_xy45i7wd.PValue, "
+              "T1.value, T2.value, T3.value "
+              "FROM (Probeset, ProbesetXRef, ProbesetFreeze) "
+              "LEFT JOIN temp_table_xy45i7wd ON temp_table_xy45i7wd.Symbol=ProbeSet.Symbol "
+              "LEFT JOIN ProbesetData AS T1 "
+              "ON T1.Id = ProbesetXRef.DataId "
+              "AND T1.StrainId=%(T1_sample_id)s "
+              "LEFT JOIN ProbesetData AS T2 "
+              "ON T2.Id = ProbesetXRef.DataId "
+              "AND T2.StrainId=%(T2_sample_id)s "
+              "LEFT JOIN ProbesetData AS T3 "
+              "ON T3.Id = ProbesetXRef.DataId "
+              "AND T3.StrainId=%(T3_sample_id)s "
+              "WHERE ProbeSet.Symbol IS NOT NULL "
+              "AND temp_table_xy45i7wd.Correlation IS NOT NULL "
+              "AND ProbesetXRef.ProbesetFreezeId = ProbesetFreeze.Id "
+              "AND ProbesetFreeze.Name = %(db_name)s "
+              "AND Probeset.Id = ProbesetXRef.ProbesetId "
+              "ORDER BY Probeset.Id"),
+             3))
diff --git a/tests/unit/db/test_traits.py b/tests/unit/db/test_traits.py
index 8af8e82..4aa9389 100644
--- a/tests/unit/db/test_traits.py
+++ b/tests/unit/db/test_traits.py
@@ -2,6 +2,8 @@
 from unittest import mock, TestCase
 from gn3.db.traits import (
     build_trait_name,
+    export_trait_data,
+    export_informative,
     set_haveinfo_field,
     update_sample_data,
     retrieve_trait_info,
@@ -12,6 +14,38 @@ from gn3.db.traits import (
     retrieve_publish_trait_info,
     retrieve_probeset_trait_info)
 
+samplelist = ["B6cC3-1", "BXD1", "BXD12", "BXD16", "BXD19", "BXD2"]
+trait_data = {
+    "mysqlid": 36688172,
+    "data": {
+        "B6cC3-1": {"sample_name": "B6cC3-1", "value": 7.51879, "variance": None, "ndata": None},
+        "BXD1": {"sample_name": "BXD1", "value": 7.77141, "variance": None, "ndata": None},
+        "BXD12": {"sample_name": "BXD12", "value": 8.39265, "variance": None, "ndata": None},
+        "BXD16": {"sample_name": "BXD16", "value": 8.17443, "variance": None, "ndata": None},
+        "BXD19": {"sample_name": "BXD19", "value": 8.30401, "variance": None, "ndata": None},
+        "BXD2": {"sample_name": "BXD2", "value": 7.80944, "variance": None, "ndata": None},
+        "BXD21": {"sample_name": "BXD21", "value": 8.93809, "variance": None, "ndata": None},
+        "BXD24": {"sample_name": "BXD24", "value": 7.99415, "variance": None, "ndata": None},
+        "BXD27": {"sample_name": "BXD27", "value": 8.12177, "variance": None, "ndata": None},
+        "BXD28": {"sample_name": "BXD28", "value": 7.67688, "variance": None, "ndata": None},
+        "BXD32": {"sample_name": "BXD32", "value": 7.79062, "variance": None, "ndata": None},
+        "BXD39": {"sample_name": "BXD39", "value": 8.27641, "variance": None, "ndata": None},
+        "BXD40": {"sample_name": "BXD40", "value": 8.18012, "variance": None, "ndata": None},
+        "BXD42": {"sample_name": "BXD42", "value": 7.82433, "variance": None, "ndata": None},
+        "BXD6": {"sample_name": "BXD6", "value": 8.09718, "variance": None, "ndata": None},
+        "BXH14": {"sample_name": "BXH14", "value": 7.97475, "variance": None, "ndata": None},
+        "BXH19": {"sample_name": "BXH19", "value": 7.67223, "variance": None, "ndata": None},
+        "BXH2": {"sample_name": "BXH2", "value": 7.93622, "variance": None, "ndata": None},
+        "BXH22": {"sample_name": "BXH22", "value": 7.43692, "variance": None, "ndata": None},
+        "BXH4": {"sample_name": "BXH4", "value": 7.96336, "variance": None, "ndata": None},
+        "BXH6": {"sample_name": "BXH6", "value": 7.75132, "variance": None, "ndata": None},
+        "BXH7": {"sample_name": "BXH7", "value": 8.12927, "variance": None, "ndata": None},
+        "BXH8": {"sample_name": "BXH8", "value": 6.77338, "variance": None, "ndata": None},
+        "BXH9": {"sample_name": "BXH9", "value": 8.03836, "variance": None, "ndata": None},
+        "C3H/HeJ": {"sample_name": "C3H/HeJ", "value": 7.42795, "variance": None, "ndata": None},
+        "C57BL/6J": {"sample_name": "C57BL/6J", "value": 7.50606, "variance": None, "ndata": None},
+        "DBA/2J": {"sample_name": "DBA/2J", "value": 7.72588, "variance": None, "ndata": None}}}
+
 class TestTraitsDBFunctions(TestCase):
     "Test cases for traits functions"
 
@@ -226,3 +260,145 @@ class TestTraitsDBFunctions(TestCase):
             with self.subTest(trait_info=trait_info, expected=expected):
                 self.assertEqual(
                     set_confidential_field(trait_type, trait_info), expected)
+
+    def test_export_trait_data_dtype(self):
+        """
+        Test `export_trait_data` with different values for the `dtype` keyword
+        argument
+        """
+        for dtype, expected in [
+                ["val", (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)],
+                ["var", (None, None, None, None, None, None)],
+                ["N", (None, None, None, None, None, None)],
+                ["all", (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)]]:
+            with self.subTest(dtype=dtype):
+                self.assertEqual(
+                    export_trait_data(trait_data, samplelist, dtype=dtype),
+                    expected)
+
+    def test_export_trait_data_dtype_all_flags(self):
+        """
+        Test `export_trait_data` with different values for the `dtype` keyword
+        argument and the different flags set up
+        """
+        for dtype, vflag, nflag, expected in [
+                ["val", False, False,
+                 (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)],
+                ["val", False, True,
+                 (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)],
+                ["val", True, False,
+                 (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)],
+                ["val", True, True,
+                 (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)],
+                ["var", False, False, (None, None, None, None, None, None)],
+                ["var", False, True, (None, None, None, None, None, None)],
+                ["var", True, False, (None, None, None, None, None, None)],
+                ["var", True, True, (None, None, None, None, None, None)],
+                ["N", False, False, (None, None, None, None, None, None)],
+                ["N", False, True, (None, None, None, None, None, None)],
+                ["N", True, False, (None, None, None, None, None, None)],
+                ["N", True, True, (None, None, None, None, None, None)],
+                ["all", False, False,
+                 (7.51879, 7.77141, 8.39265, 8.17443, 8.30401, 7.80944)],
+                ["all", False, True,
+                 (7.51879, None, 7.77141, None, 8.39265, None, 8.17443, None,
+                  8.30401, None, 7.80944, None)],
+                ["all", True, False,
+                 (7.51879, None, 7.77141, None, 8.39265, None, 8.17443, None,
+                  8.30401, None, 7.80944, None)],
+                ["all", True, True,
+                 (7.51879, None, None, 7.77141, None, None, 8.39265, None, None,
+                  8.17443, None, None, 8.30401, None, None, 7.80944, None, None)]
+        ]:
+            with self.subTest(dtype=dtype, vflag=vflag, nflag=nflag):
+                self.assertEqual(
+                    export_trait_data(
+                        trait_data, samplelist, dtype=dtype, var_exists=vflag,
+                        n_exists=nflag),
+                    expected)
+
+    def test_export_informative(self):
+        """Test that the function exports appropriate data."""
+        # pylint: disable=W0621
+        for trait_data, inc_var, expected in [
+                [{"data": {
+                    "sample1": {
+                        "sample_name": "sample1", "value": 9, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample2": {
+                        "sample_name": "sample2", "value": 8, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample3": {
+                        "sample_name": "sample3", "value": 7, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample4": {
+                        "sample_name": "sample4", "value": 6, "variance": None,
+                        "ndata": 13
+                    },
+                }}, 0, (
+                    ("sample1", "sample2", "sample3", "sample4"), (9, 8, 7, 6),
+                    (None, None, None, None))],
+                [{"data": {
+                    "sample1": {
+                        "sample_name": "sample1", "value": 9, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample2": {
+                        "sample_name": "sample2", "value": 8, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample3": {
+                        "sample_name": "sample3", "value": None, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample4": {
+                        "sample_name": "sample4", "value": 6, "variance": None,
+                        "ndata": 13
+                    },
+                }}, 0, (
+                    ("sample1", "sample2", "sample4"), (9, 8, 6),
+                    (None, None, None))],
+                [{"data": {
+                    "sample1": {
+                        "sample_name": "sample1", "value": 9, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample2": {
+                        "sample_name": "sample2", "value": 8, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample3": {
+                        "sample_name": "sample3", "value": 7, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample4": {
+                        "sample_name": "sample4", "value": 6, "variance": None,
+                        "ndata": 13
+                    },
+                }}, True, (tuple(), tuple(), tuple())],
+                [{"data": {
+                    "sample1": {
+                        "sample_name": "sample1", "value": 9, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample2": {
+                        "sample_name": "sample2", "value": 8, "variance": 0.657,
+                        "ndata": 13
+                    },
+                    "sample3": {
+                        "sample_name": "sample3", "value": 7, "variance": None,
+                        "ndata": 13
+                    },
+                    "sample4": {
+                        "sample_name": "sample4", "value": 6, "variance": None,
+                        "ndata": 13
+                    },
+                }}, 0, (
+                    ("sample1", "sample2", "sample3", "sample4"), (9, 8, 7, 6),
+                    (None, 0.657, None, None))]]:
+            with self.subTest(trait_data=trait_data):
+                self.assertEqual(
+                    export_informative(trait_data, inc_var), expected)