Kl/validty updates

gnomad_qc/federated/__init__.py

@@ -0,0 +1 @@
+# noqa: D104X

gnomad_qc/v5/configs/validity_inputs_config.json → gnomad_qc/federated/configs/validity_inputs_config.json

@@ -1,12 +1,12 @@
 {
-    "missingness_threshold": 0.5,
-    "struct_annotations_for_missingness": [],
+    "struct_annotations_to_skip_missingness": ["vep"],
     "freq_fields": {"freq": "freq", "freq_meta": "freq_meta", "freq_meta_sample_count": "freq_meta_sample_count"},
     "faf_fields": {},
     "freq_annotations_to_sum": ["AC", "AN", "homozygote_count"],
     "sort_order": ["subset", "gen_anc", "sex", "group"],
     "nhomalt_metric": "homozygote_count",
     "subsets": [""],
     "variant_filter_field": "",
-    "check_mono_and_only_het" : true
+    "data_type": "genomes",
+    "check_mono_and_only_het": true
 }

gnomad_qc/v5/configs/validity_inputs_schema.py → gnomad_qc/federated/configs/validity_inputs_schema.py

@@ -3,7 +3,10 @@
 schema = {
     "type": "object",
     "properties": {
-        "missingness_threshold": {"type": "number"},
+        "struct_annotations_to_skip_missingness": {
+            "type": "array",
+            "items": {"type": "string"},
+        },
         "struct_annotations_for_missingness": {
             "type": "array",
             "items": {"type": "string"},
@@ -46,17 +49,18 @@
             "items": {"type": "string"},
         },
         "variant_filter_field": {"type": "string"},
+        "data_type": {"type": "string", "enum": ["exomes", "genomes"]},
         "check_mono_and_only_het": {"type": "boolean"},
     },
     "required": [
-        "missingness_threshold",
         "freq_fields",
         "freq_annotations_to_sum",
         "sort_order",
         "nhomalt_metric",
         "subsets",
         "variant_filter_field",
         "check_mono_and_only_het",
+        "data_type",
     ],
     "additionalProperties": False,
 }