Fix missing hgmd variants (#747)

* Fix missing recoding

* Fix hgmd parsing

* lint

* Update tables

* Update versions

Author: bpblanken

v03_pipeline/lib/reference_data/config.py

@@ -239,6 +239,7 @@ def custom_mpc_select(ht):
     'hgmd': {
         '37': {
             'custom_import': download_and_import_hgmd_vcf,
+            'version': 'HGMD_Pro_2023',
             'source_path': 'gs://seqr-reference-data-private/GRCh37/HGMD/HGMD_Pro_2023.1_hg19.vcf.gz',
             'select': {'accession': 'rsid', 'class': 'info.CLASS'},
             'enum_select': {
@@ -254,6 +255,7 @@ def custom_mpc_select(ht):
         },
         '38': {
             'custom_import': download_and_import_hgmd_vcf,
+            'version': 'HGMD_Pro_2023',
             'source_path': 'gs://seqr-reference-data-private/GRCh38/HGMD/HGMD_Pro_2023.1_hg38.vcf.gz',
             'select': {'accession': 'rsid', 'class': 'info.CLASS'},
             'enum_select': {

v03_pipeline/lib/reference_data/hgmd.py

@@ -13,5 +13,6 @@ def download_and_import_hgmd_vcf(
         force=True,
         min_partitions=100,
         skip_invalid_loci=True,
+        contig_recoding=reference_genome.contig_recoding(),
     )
     return mt.rows()

v03_pipeline/lib/reference_data/hgmd_test.py

@@ -0,0 +1,12 @@
+import unittest
+
+from v03_pipeline.lib.model import ReferenceGenome
+from v03_pipeline.lib.reference_data.hgmd import download_and_import_hgmd_vcf
+
+TEST_HGMD_VCF = 'v03_pipeline/var/test/reference_data/test_hgmd.vcf'
+
+
+class HGMDTest(unittest.TestCase):
+    def test_import_hgmd_vcf(self):
+        ht = download_and_import_hgmd_vcf(TEST_HGMD_VCF, ReferenceGenome.GRCh38)
+        self.assertEqual(ht.count(), 1)

v03_pipeline/lib/tasks/reference_data/update_variant_annotations_table_with_updated_reference_dataset_test.py

@@ -789,7 +789,7 @@ def test_update_vat_with_updated_rdc_snv_indel_38(
                         topmed=None,
                         gnomad_non_coding_constraint=None,
                         screen=None,
-                        hgmd=None,
+                        hgmd='HGMD_Pro_2023',
                     ),
                     enums=hl.Struct(
                         cadd=hl.Struct(),
@@ -1053,7 +1053,7 @@ def test_update_vat_with_updated_rdc_snv_indel_37(
                         primate_ai='v0.2',
                         splice_ai=None,
                         topmed=None,
-                        hgmd=None,
+                        hgmd='HGMD_Pro_2023',
                     ),
                     enums=hl.Struct(
                         cadd=hl.Struct(),

v03_pipeline/lib/tasks/update_variant_annotations_table_with_new_samples_test.py

@@ -472,7 +472,7 @@ def test_multiple_update_vat(
                         topmed=None,
                         gnomad_non_coding_constraint=None,
                         screen=None,
-                        hgmd=None,
+                        hgmd='HGMD_Pro_2023',
                     ),
                     enums=hl.Struct(
                         cadd=hl.Struct(),

v03_pipeline/var/test/reference_data/test_hgmd.vcf

@@ -0,0 +1,20 @@
+##fileformat=VCFv4.2
+##note=VCF file is compatible with VCFv4.3 if required by updating fileformat parameter to v4.3
+##source=HGMD_PRO_2023.1
+##reference=GRCh38
+##comment="REF and ALT sequences are both on forward strand of reference assembly"
+##INFO=<ID=CLASS,Number=1,Type=String>
+##INFO=<ID=MUT,Number=1,Type=String>
+##INFO=<ID=GENE,Number=1,Type=String>
+##INFO=<ID=STRAND,Number=1,Type=String>
+##INFO=<ID=DNA,Number=1,Type=String>
+##INFO=<ID=PROT,Number=1,Type=String>
+##INFO=<ID=DB,Number=1,Type=String>
+##INFO=<ID=PHEN,Number=1,Type=String>
+##INFO=<ID=RANKSCORE,Number=1,Type=Float>
+##INFO=<ID=SVTYPE,Number=1,Type=String>
+##INFO=<ID=END,Number=1,Type=Integer>
+##INFO=<ID=SVLEN,Number=1,Type=Integer>
+##ALT=<ID=DEL>
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+1	925942	CM2039807	A	G	.	.	CLASS=DM?;MUT=ALT;GENE=SAMD11;STRAND=+;DNA=NM_152486.4%3Ac.1A>G;PROT=NP_689699.3%3Ap.M1?;DB=rs1403306757;PHEN="Retinitis_pigmentosa";RANKSCORE=0.57

v03_pipeline/var/test/reference_data/test_hgmd_1.ht/.README.txt.crc

@@ -1,3 +1,3 @@
 This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
-  Written with version 0.2.128-eead8100a1c1
-  Created at 2024/03/11 13:33:57
+  Written with version 0.2.120-f00f916faf78
+  Created at 2024/03/28 16:27:47