Bump a few versions

Author: bpblanken

v03_pipeline/lib/misc/nested_field.py

@@ -12,4 +12,7 @@ def parse_nested_field(t: hl.MatrixTable | hl.Table, fields: str):
             ]
         else:
             expression = expression[field]
+    # Parse float64s into float32s to save space!
+    if expression.dtype == hl.tfloat64:
+        expression = hl.float32(expression)
     return expression

v03_pipeline/lib/reference_datasets/eigen.py

@@ -3,5 +3,5 @@
 
 def get_ht(path: str, *_) -> hl.Table:
     ht = hl.read_table(path)
-    ht = ht.select(Eigen_phred=ht.info['Eigen-phred'])
+    ht = ht.select(Eigen_phred=hl.float32(ht.info['Eigen-phred']))
     return ht.group_by(*ht.key).aggregate(Eigen_phred=hl.agg.max(ht.Eigen_phred))

v03_pipeline/lib/reference_datasets/exac.py

@@ -5,7 +5,6 @@
 from v03_pipeline.lib.reference_datasets.misc import vcf_to_ht
 
 SELECT = {
-    'AF_POPMAX': 'info.POPMAX',
     'AF': 'info.AF#',
     'AC_Adj': 'info.AC_Adj#',
     'AC_Het': 'info.AC_Het#',
@@ -18,5 +17,12 @@
 def get_ht(path: str, reference_genome: ReferenceGenome) -> hl.Table:
     ht = vcf_to_ht(path, reference_genome, split_multi=True)
     return ht.select(
+        AF_POPMAX=hl.or_missing(
+            ht.info.AC_POPMAX[ht.a_index - 1] != 'NA',
+            hl.float32(
+                hl.int32(ht.info.AC_POPMAX[ht.a_index - 1])
+                / hl.int32(ht.info.AN_POPMAX[ht.a_index - 1]),
+            ),
+        ),
         **{k: parse_nested_field(ht, v) for k, v in SELECT.items()},
     )

v03_pipeline/lib/reference_datasets/hmtvar.py

@@ -19,7 +19,7 @@ def get_ht(
             reference_genome.value,
         ),
         alleles=hl.array([ht.ref_rCRS, ht.alt]),
-        score=ht.disease_score,
+        score=hl.float32(ht.disease_score),
     )
     ht = ht.key_by('locus', 'alleles')
     ht = ht.filter(

v03_pipeline/lib/reference_datasets/reference_dataset.py

@@ -204,14 +204,14 @@ def get_ht(
     ReferenceDataset.eigen: {
         ReferenceGenome.GRCh37: {
             DATASET_TYPES: frozenset([DatasetType.SNV_INDEL]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             # NB: The download link on the Eigen website (http://www.columbia.edu/~ii2135/download.html) is broken
             # as of 11/15/24 so we will host the data
             PATH: 'gs://seqr-reference-data/GRCh37/eigen/EIGEN_coding_noncoding.grch37.ht',
         },
         ReferenceGenome.GRCh38: {
             DATASET_TYPES: frozenset([DatasetType.SNV_INDEL]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             PATH: 'gs://seqr-reference-data/GRCh38/eigen/EIGEN_coding_noncoding.liftover_grch38.ht',
         },
     },
@@ -232,12 +232,12 @@ def get_ht(
     ReferenceDataset.exac: {
         ReferenceGenome.GRCh37: {
             DATASET_TYPES: frozenset([DatasetType.SNV_INDEL]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             PATH: 'gs://gcp-public-data--gnomad/legacy/exacv1_downloads/release1/ExAC.r1.sites.vep.vcf.gz',
         },
         ReferenceGenome.GRCh38: {
             DATASET_TYPES: frozenset([DatasetType.SNV_INDEL]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             # NB: Exac is only available on GRCh37 so we host a lifted over version
             PATH: 'gs://seqr-reference-data/GRCh38/gnomad/ExAC.r1.sites.liftover.b38.vcf.gz',
         },
@@ -294,7 +294,7 @@ def get_ht(
     ReferenceDataset.hmtvar: {
         ReferenceGenome.GRCh38: {
             DATASET_TYPES: frozenset([DatasetType.MITO]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             #  NB: https://www.hmtvar.uniba.it is unavailable as of 11/15/24 so we will host the data
             PATH: 'https://storage.googleapis.com/seqr-reference-data/GRCh38/mitochondrial/HmtVar/HmtVar%20Jan.%2010%202022.json',
         },

v03_pipeline/var/test/reference_datasets/GRCh37/eigen/1.1.ht/.README.txt.crc

@@ -0,0 +1,3 @@
+This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
+  Written with version 0.2.133-4c60fddb171a
+  Created at 2024/12/02 19:45:01