Merge pull request #67 from Sage-Bionetworks-Workflows/DPE-1445

[DPE 1445] Add HTAN H5AD validation tool

Author: andrewelamb

setup.cfg

@@ -4,7 +4,7 @@
 # https://setuptools.pypa.io/en/latest/references/keywords.html
 
 [metadata]
-version = 1.7.5
+version = 1.8.0
 name = dcqc
 description = Python package for performing quality control (QC) for data coordination (DC)
 author = Bruno Grande

src/dcqc/suites/suites.py

@@ -86,3 +86,4 @@ class H5ADSuite(HDF5Suite):
     """
 
     file_type = FileType.get_file_type("H5AD")
+    add_tests = (tests.H5adHtanValidatorTest,)

src/dcqc/tests/__init__.py

@@ -8,6 +8,7 @@
 from dcqc.tests.bioformats_info_test import BioFormatsInfoTest
 from dcqc.tests.file_extension_test import FileExtensionTest
 from dcqc.tests.grep_date_test import GrepDateTest
+from dcqc.tests.h5ad_htan_validator_test import H5adHtanValidatorTest
 from dcqc.tests.json_load_test import JsonLoadTest
 from dcqc.tests.jsonld_load_test import JsonLdLoadTest
 from dcqc.tests.libtiff_info_test import LibTiffInfoTest

src/dcqc/tests/h5ad_htan_validator_test.py

@@ -0,0 +1,29 @@
+from dcqc.target import SingleTarget
+from dcqc.tests.base_test import ExternalBaseTest, Process, TestTier
+
+
+class H5adHtanValidatorTest(ExternalBaseTest):
+    """
+    Based on [HTAN H5AD Validator](https://github.com/ncihtan/h5ad)
+    This is an h5ad validator for HTAN Phase 2 single
+     cell/single nuclei RNA-sequencing data.
+    """
+
+    tier = TestTier.INTERNAL_CONFORMANCE
+    pass_code = 0
+    fail_code = 1
+    failure_reason_location = "std_out"
+    target: SingleTarget
+
+    def generate_process(self) -> Process:
+        path = self.target.file.stage()
+        command_args = [
+            "python",
+            "/usr/local/bin/h5ad.py",
+            f"'{path.name}'",
+        ]
+        process = Process(
+            container="ghcr.io/sage-bionetworks-workflows/htan-h5ad-validator:0.1.1",
+            command_args=command_args,
+        )
+        return process

tests/conftest.py

@@ -65,6 +65,8 @@ def test_files(get_data):
     tiff_dirty_datetime_path = get_data("test_image_dirty_datetime.tif")
     tiff_date_in_tag_path = get_data("date_tag.tif")
     invalid_xml_ome_tiff_path = get_data("invalid_xml.ome.tif")
+    htan_good_h5ad_path = get_data("htan_good.h5ad")
+    htan_bad_h5ad_path = get_data("htan_bad.h5ad")
     invalid_xml_metadata = {
         "file_type": "tiff",
         "md5_checksum": "a2550a887091d51351d547c8beae8f0c",
@@ -98,6 +100,14 @@ def test_files(get_data):
         "file_type": "tiff",
         "md5_checksum": "28a9ee7d0e994d494068ce8d6cda0268",
     }
+    htan_good_h5ad_metadata = {
+        "file_type": "h5ad",
+        "md5_checksum": "ba11278e2c19b10851a66587e17ec97f",
+    }
+    htan_bad_h5ad_metadata = {
+        "file_type": "h5ad",
+        "md5_checksum": "e6f5e1a812c699af489352652cf0238d",
+    }
     test_files = {
         "date_in_tag_tiff": File(tiff_date_in_tag_path.as_posix(), tiff_metadata),
         "good_txt": File(txt_path.as_posix(), good_metadata),
@@ -115,6 +125,8 @@ def test_files(get_data):
         "dirty_datetime_in_tag_tiff": File(
             tiff_dirty_datetime_path.as_posix(), tiff_dirty_datetime_metadata
         ),
+        "htan_good_h5ad": File(htan_good_h5ad_path.as_posix(), htan_good_h5ad_metadata),
+        "htan_bad_h5ad": File(htan_bad_h5ad_path.as_posix(), htan_bad_h5ad_metadata),
     }
 
     # Create an in-memory remote file based on the good file

tests/data/htan_bad.h5ad

@@ -437,3 +437,85 @@ def test_that_the_tiffdatetimetest_correctly_interprets_exit_code_0_and_1(
             )
             test_status = test.get_status()
             assert test_status == TestStatus.FAIL
+
+
+class TestH5adHtanValidatorTest:
+    @pytest.fixture(scope="function", autouse=True)
+    def setup_method(self, test_targets):
+        self.good_h5ad_target = test_targets["htan_good_h5ad"]
+        self.good_h5ad_test = tests.H5adHtanValidatorTest(self.good_h5ad_target)
+        self.bad_h5ad_target = test_targets["htan_bad_h5ad"]
+        self.bad_h5ad_test = tests.H5adHtanValidatorTest(self.bad_h5ad_target)
+        self.txt_target = test_targets["good_txt"]
+        self.txt_test = tests.H5adHtanValidatorTest(self.txt_target)
+
+    def test_that_the_command_is_produced(self):
+        process = self.good_h5ad_test.generate_process()
+        assert "/usr/local/bin/h5ad.py" in process.command
+
+    @docker_enabled_test
+    def test_that_the_exit_code_is_1_when_it_should_be(self):
+        process = self.bad_h5ad_test.generate_process()
+        executor = DockerExecutor(
+            process.container, process.command, self.bad_h5ad_target.file.url
+        )
+        executor.execute()
+        assert "File:  htan_bad.h5ad" in executor.std_out
+        assert "Cellxgene run has errors. " in executor.std_out
+        assert "HTAN Validation Failed." in executor.std_out
+        assert "'cellxgene-schema output: Starting validation" in executor.std_out
+        assert "HTAN-specific Validation Errors:" in executor.std_out
+        assert executor.exit_code == "1"
+
+    @docker_enabled_test
+    def test_that_the_exit_code_is_0_when_it_should_be(self):
+        process = self.good_h5ad_test.generate_process()
+        executor = DockerExecutor(
+            process.container, process.command, self.good_h5ad_target.file.url
+        )
+        executor.execute()
+        assert "File:  htan_good.h5ad" in executor.std_out
+        assert "Cellxgene run successful." in executor.std_out
+        assert "Validation Passed!" in executor.std_out
+        assert executor.exit_code == "0"
+
+    @docker_enabled_test
+    def test_that_wrong_file_type_has_1_exit_code(self):
+        process = self.txt_test.generate_process()
+        executor = DockerExecutor(
+            process.container, process.command, self.txt_target.file.url
+        )
+        executor.execute()
+        print(executor.std_out)
+        assert executor.std_out == (
+            "HTAN h5ad File Validator\n"
+            "File:  test.txt\n"
+            "An error occurred while trying to open test.txt\n"
+            "Unable to synchronously open file (file signature not found)\n"
+            "\n"
+        )
+        assert executor.exit_code == "1"
+
+    def test_that_the_test_correctly_interprets_exit_code_0_and_1(self, mocker):
+        # 1 is pass, 0 is fail
+        with TemporaryDirectory() as tmp_dir:
+            path_0 = Path(tmp_dir, "code_0.txt")
+            path_1 = Path(tmp_dir, "code_1.txt")
+            path_0.write_text("0")
+            path_1.write_text("1")
+            fail_outputs = {"std_out": path_1, "std_err": path_1, "exit_code": path_0}
+            pass_outputs = {"std_out": path_0, "std_err": path_0, "exit_code": path_1}
+
+            test = tests.TiffDateTimeTest(self.good_h5ad_target)
+            mocker.patch.object(
+                test, "_find_process_outputs", return_value=pass_outputs
+            )
+            test_status = test.get_status()
+            assert test_status == TestStatus.PASS
+
+            test = tests.TiffDateTimeTest(self.bad_h5ad_target)
+            mocker.patch.object(
+                test, "_find_process_outputs", return_value=fail_outputs
+            )
+            test_status = test.get_status()
+            assert test_status == TestStatus.FAIL