Add code to check genome section in MANIFEST before loading sample

This includes tests to cover the different cases where the genome
section of the MANIFEST is invalid.

Signed-off-by: Don Naegely <naegelyd@gmail.com>

tests/cases/sample_load_process/tests.py

@@ -153,6 +153,91 @@ def test_pipeline(self):
         system = System.get(Sample.objects.all()[0])
         self.assertEqual(len(system), 3)
 
+    @override_settings(REQUIRED_GENOME_VERSION='hg19')
+    def test_missing_genome_section(self):
+        expected_counts = {
+            'batches': 1,
+            'results_per_sample': [
+                {
+                    'batch': 'batch1',
+                    'sample': 'NA12891',
+                    'count': 1963,
+                },
+                {
+                    'batch': 'batch1',
+                    'sample': 'NA12892',
+                    'count': 1963,
+                },
+                {
+                    'batch': 'batch1',
+                    'sample': 'NA12878',
+                    'count': 1963,
+                }
+            ],
+            'samples': 3,
+            'samples_per_batch': [(1, 3)],
+        }
+        expected_counts['results'] = \
+            sum([x['count'] for x in expected_counts['results_per_sample']])
+
+        # Immediately validates and creates a sample.
+        management.call_command('samples', 'queue')
+
+        # Synchronously work on queue.
+        worker1 = get_worker('variants')
+        worker2 = get_worker('default')
+
+        # Work on variants.
+        worker1.work(burst=True)
+
+        # Work on effects.
+        worker2.work(burst=True)
+
+        # Since the MANIFEST for Batch 2 has no genome version listed, we
+        # should only have data for samples in Batch 1. Perform all the checks
+        # against our trimmed list of expected counts.
+        self.assertEqual(Result.objects.count(), expected_counts['results'])
+
+        # Batches are now published..
+        self.assertEqual(Batch.objects.filter(published=True).count(),
+                         expected_counts['batches'])
+
+        # Ensure the counts are accurate for each sample..
+        for ec in expected_counts['results_per_sample']:
+            sample = Sample.objects.get(name=ec['sample'],
+                                        batch__name=ec['batch'])
+            self.assertTrue(sample.published)
+            self.assertEqual(sample.count, ec['count'])
+
+        # Batches are created with the samples, but are unpublished
+        for pk, count in expected_counts['samples_per_batch']:
+            batch = Batch.objects.get(pk=pk)
+            self.assertTrue(batch.published)
+            self.assertEqual(batch.count, count)
+
+    @override_settings(REQUIRED_GENOME_VERSION='hg18')
+    def test_wrong_genome_version(self):
+        # Immediately validates and creates a sample.
+        management.call_command('samples', 'queue')
+
+        # Synchronously work on queue.
+        worker1 = get_worker('variants')
+        worker2 = get_worker('default')
+
+        # Work on variants.
+        worker1.work(burst=True)
+
+        # Work on effects.
+        worker2.work(burst=True)
+
+        # Since the genome version was required but does not match any of the
+        # versions specified in the MANIFESTs, we should have no data.
+        self.assertEqual(Variant.objects.count(), 0)
+        self.assertEqual(Sample.objects.count(), 0)
+        self.assertEqual(Cohort.objects.count(), 0)
+        self.assertEqual(Batch.objects.count(), 0)
+        self.assertEqual(Project.objects.count(), 0)
+
 
 class SnpeffReloadTest(QueueTestCase):
     def test(self):

tests/samples/batch1/locus_1/MANIFEST

@@ -9,3 +9,7 @@ version = 0
 [vcf]
 file = locus_1.vcf
 md5 = f67a6913ee83345657a6e790c6f5feee
+
+[genome]
+name = Genome Reference Consortium GRCh37
+version = hg19

vdw/samples/pipeline/handlers.py

@@ -3,6 +3,7 @@
 import sts
 import time
 import vcf
+from django.conf import settings
 from django.db import connections, transaction
 from sts.contextmanagers import transition
 from vdw.pipeline import checks, job, ManifestReader
@@ -51,26 +52,39 @@ def check_sample_section(manifest):
 def load_samples(manifest_path, database, **kwargs):
     manifest = ManifestReader(manifest_path)
 
-    # Ensure the sample is marked to be loaded..
+    # If a specific version of the genome is required then make sure the
+    # MANIFEST lists that version or abort the load process.
+    genome_version = getattr(settings, 'REQUIRED_GENOME_VERSION', None)
+    if genome_version:
+        genome_info = manifest.section('genome')
+
+        if not genome_info:
+            log.info('Genome version "{0}" is required but genome section is '
+                     'not included in MANIFEST({1}). Skipping sample.'
+                     .format(genome_version, manifest_path))
+            return
+
+        if genome_version != genome_info['version']:
+            log.info(
+                'Genome version "{0}" is required but version "{1}" was found '
+                'in the MANIFEST({2}). Skipping sample.'
+                .format(genome_version, genome_info['version'], manifest_path))
+            return
+
+    # Ensure the sample is marked to be loaded.
     if not manifest.marked_for_load():
         log.info('Sample not marked for load', extra={
             'manifest_path': manifest_path,
         })
         return
 
-    # Ensure the sample section is valid..
+    # Ensure the sample section is valid.
     if not check_sample_section(manifest):
         log.info('Manifest sample section is not valid', extra={
             'manifest_path': manifest_path,
         })
         return
 
-    # [sample]
-    # project = PCGC
-    # batch = OTHER
-    # sample = 1-03131
-    # version = 1
-
     sample_info = manifest.section('sample')
     vcf_info = manifest.section('vcf')