array

Author: jklugherz

v03_pipeline/lib/methods/sample_qc.py

@@ -57,12 +57,11 @@ def annotate_filter_flags(
     else:
         flags['coverage'] = mt.mean_coverage < WGS_CALLRATE_LOW_THRESHOLD
 
-    mt = mt.annotate_cols(
-        filter_flags=hl.set(
+    return mt.annotate_cols(
+        filter_flags=hl.array(
             [hl.or_missing(filter_cond, name) for name, filter_cond in flags.items()],
         ).filter(hl.is_defined),
-    )
-    return mt.drop(
+    ).drop(
         'contamination_rate',
         'percent_bases_at_20x',
         'mean_coverage',

v03_pipeline/lib/tasks/write_sample_qc_json.py

@@ -54,9 +54,7 @@ def run(self):
             r = dict(row)
             sample_id = r.pop('s')
             for field, value in r.items():
-                sample_qc_dict[sample_id][field] = (
-                    list(value) if isinstance(value, set) else value
-                )
+                sample_qc_dict[sample_id][field] = value
 
         with self.output().open('w') as f:
             json.dump(sample_qc_dict, f)

v03_pipeline/lib/tasks/write_sample_qc_json_test.py

@@ -103,30 +103,26 @@ def test_call_sample_qc(
         self.assertTrue(hfs.exists(task.output().path))
 
         with task.output().open('r') as f:
+            res = json.load(f)
             self.assertCountEqual(
-                json.load(f),
+                res['HG00731'],
                 {
-                    'HG00731': {
-                        'sample_type': 'WGS',
-                        'filter_flags': ['contamination', 'coverage'],
-                    },
-                    'HG00732': {'sample_type': 'WGS', 'filter_flags': ['coverage']},
-                    'HG00733': {
-                        'sample_type': 'WGS',
-                        'filter_flags': ['contamination'],
-                    },
-                    'NA19675': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA19678': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA19679': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20870': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20872': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20874': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20875': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20876': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20877': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20878': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20881': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20885': {'sample_type': 'WGS', 'filter_flags': []},
-                    'NA20888': {'sample_type': 'WGS', 'filter_flags': []},
+                    'sample_type': 'WGS',
+                    'filter_flags': ['contamination', 'coverage'],
                 },
             )
+            self.assertCountEqual(
+                res['HG00732'],
+                {'sample_type': 'WGS', 'filter_flags': ['coverage']},
+            )
+            self.assertCountEqual(
+                res['HG00733'],
+                {
+                    'sample_type': 'WGS',
+                    'filter_flags': ['contamination'],
+                },
+            )
+            self.assertCountEqual(
+                res['HG00732'],
+                {'sample_type': 'WGS', 'filter_flags': []},
+            )