add apogee2 score (#955)

* add apogee2 score

* update mitimpact script

* fix other update vat test

* score

Author: jklugherz

download_and_create_reference_datasets/v02/mito/write_mito_mitimpact_ht.py

@@ -3,13 +3,13 @@
 from download_and_create_reference_datasets.v02.mito.utils import load
 
 CONFIG = {
-    'input_path': 'https://mitimpact.css-mendel.it/cdn/MitImpact_db_3.0.7.txt.zip',
+    'input_path': 'https://mitimpact.css-mendel.it/cdn/MitImpact_db_3.1.3.txt.zip',
     'input_type': 'tsv',
-    'output_path': 'gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.0.7.ht',
+    'output_path': 'gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.1.3.ht',
     'annotate': {
         'locus': lambda ht: hl.locus('chrM', hl.parse_int32(ht.Start)),
         'alleles': lambda ht: [ht.Ref, ht.Alt],
-        'APOGEE_score': lambda ht: hl.parse_float(ht.APOGEE_score),
+        'APOGEE2_score': lambda ht: hl.parse_float(ht.APOGEE2_score),
     },
 }
 

v03_pipeline/lib/reference_data/config.py

@@ -503,9 +503,9 @@ def custom_mpc_select(ht):
     },
     'mitimpact': {
         '38': {
-            'version': '3.0.7',
-            'path': 'gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.0.7.ht',
-            'select': {'score': 'APOGEE_score'},
+            'version': '3.1.3',
+            'path': 'gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.1.3.ht',
+            'select': {'score': 'APOGEE2_score'},
         },
     },
     'hmtvar': {

v03_pipeline/lib/tasks/reference_data/update_variant_annotations_table_with_updated_reference_dataset_test.py

@@ -604,13 +604,13 @@
     'mitimpact': {
         '38': {
             **CONFIG['mitimpact']['38'],
-            'source_path': 'gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.0.7.ht',
+            'source_path': 'gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.1.3.ht',
             'custom_import': lambda *_: hl.Table.parallelize(
                 [],
                 hl.tstruct(
                     locus=hl.tlocus('GRCh38'),
                     alleles=hl.tarray(hl.tstr),
-                    APOGEE_score=hl.tfloat64,
+                    APOGEE2_score=hl.tfloat64,
                 ),
                 key=['locus', 'alleles'],
                 globals=hl.Struct(),
@@ -1008,7 +1008,7 @@ def test_update_vat_with_updated_rdc_mito_38(
                         helix_mito='gs://seqr-reference-data/GRCh38/mitochondrial/Helix/HelixMTdb_20200327.ht',
                         hmtvar='gs://seqr-reference-data/GRCh38/mitochondrial/HmtVar/HmtVar%20Jan.%2010%202022.ht',
                         mitomap='gs://seqr-reference-data/GRCh38/mitochondrial/MITOMAP/mitomap-confirmed-mutations-2022-02-04.ht',
-                        mitimpact='gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.0.7.ht',
+                        mitimpact='gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.1.3.ht',
                         clinvar_mito='https://ftp.ncbi.nlm.nih.gov/pub/clinvar/vcf_GRCh38/clinvar.vcf.gz',
                         dbnsfp_mito='gs://seqr-reference-data/GRCh38/dbNSFP/v4.2/dbNSFP4.2a_variant.with_new_scores.ht',
                         high_constraint_region_mito='gs://seqr-reference-data/GRCh38/mitochondrial/Helix high constraint intervals Feb-15-2022.tsv',
@@ -1019,7 +1019,7 @@ def test_update_vat_with_updated_rdc_mito_38(
                         helix_mito='20200327',
                         hmtvar='Jan. 10 2022',
                         mitomap='Feb. 04 2022',
-                        mitimpact='3.0.7',
+                        mitimpact='3.1.3',
                         clinvar_mito='2023-07-22',
                         dbnsfp_mito='4.2',
                         high_constraint_region_mito='Feb-15-2022',
@@ -1095,7 +1095,7 @@ def test_update_vat_with_updated_rdc_mito_38(
                     ),
                     hmtvar=hl.Struct(score=0.6700000166893005),
                     mitomap=None,
-                    mitimpact=hl.Struct(score=0.5199999809265137),
+                    mitimpact=hl.Struct(score=0.42500001192092896),
                     high_constraint_region_mito=True,
                     local_constraint_mito=hl.Struct(score=0.5),
                 ),

v03_pipeline/lib/tasks/update_variant_annotations_table_with_new_samples_test.py

@@ -971,7 +971,7 @@ def test_mito_update_vat(
                         helix_mito='gs://seqr-reference-data/GRCh38/mitochondrial/Helix/HelixMTdb_20200327.ht',
                         hmtvar='gs://seqr-reference-data/GRCh38/mitochondrial/HmtVar/HmtVar%20Jan.%2010%202022.ht',
                         mitomap='gs://seqr-reference-data/GRCh38/mitochondrial/MITOMAP/mitomap-confirmed-mutations-2022-02-04.ht',
-                        mitimpact='gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.0.7.ht',
+                        mitimpact='gs://seqr-reference-data/GRCh38/mitochondrial/MitImpact/MitImpact_db_3.1.3.ht',
                         local_constraint_mito='gs://seqr-reference-data/GRCh38/mitochondrial/local_constraint.tsv',
                     ),
                     versions=hl.Struct(
@@ -982,7 +982,7 @@ def test_mito_update_vat(
                         helix_mito='20200327',
                         hmtvar='Jan. 10 2022',
                         mitomap='Feb. 04 2022',
-                        mitimpact='3.0.7',
+                        mitimpact='3.1.3',
                         local_constraint_mito='2024-07-24',
                     ),
                     enums=hl.Struct(

v03_pipeline/var/test/reference_data/test_combined_mito_1.ht/.README.txt.crc

@@ -1,3 +1,3 @@
 This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
-  Written with version 0.2.133-4c60fddb171a
-  Created at 2024/11/02 15:10:48
+  Written with version 0.2.132-678e1f52b999
+  Created at 2024/11/07 10:28:31