Merge pull request #4524 from broadinstitute/lookup-mulit-build

Variant Lookup multi genome build build

hail_search/fixtures/GRCh37/SNV_INDEL/lookup.ht/README.txt

@@ -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 2024/08/16 15:39:04
+  Created at 2025/01/02 17:23:59

hail_search/fixtures/GRCh37/SNV_INDEL/projects/WGS/R0004_non_analyst_project.ht/README.txt

@@ -0,0 +1,3 @@
+This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
+  Written with version 0.2.128-eead8100a1c1
+  Created at 2025/01/02 17:57:36

hail_search/queries/base.py

@@ -1222,7 +1222,7 @@ def gene_counts(self):
     def _filter_variant_ids(self, ht, variant_ids):
         return ht
 
-    def lookup_variants(self, variant_ids):
+    def lookup_variants(self, variant_ids, additional_annotations=None):
         self._parse_intervals(intervals=None, variant_ids=variant_ids, variant_keys=variant_ids)
         ht = self._read_table('annotations.ht', drop_globals=['versions'])
         ht = self._filter_variant_ids(ht, variant_ids)
@@ -1232,6 +1232,8 @@ def lookup_variants(self, variant_ids):
             k: v for k, v in self.annotation_fields(include_genotype_overrides=False).items()
             if k not in {FAMILY_GUID_FIELD, GENOTYPES_FIELD}
         }
+        if additional_annotations:
+            annotation_fields.update(additional_annotations)
         formatted = self._format_results(ht.key_by(), annotation_fields=annotation_fields, include_genotype_overrides=False)
 
         return formatted.aggregate(hl.agg.take(formatted.row, len(variant_ids)))
@@ -1246,9 +1248,13 @@ def _get_variant_project_data(self, variant_id, **kwargs):
         return project_data[0] if project_data else {}
 
     def lookup_variant(self, variant_id, **kwargs):
-        variants = self.lookup_variants([variant_id])
+        variants = self.lookup_variants([variant_id], additional_annotations=self._lookup_variant_annotations())
         if not variants:
             raise HTTPNotFound()
         variant = dict(variants[0])
-        variant.update(self._get_variant_project_data(variant_id, **kwargs))
+        variant.update(self._get_variant_project_data(variant_id, variant=variant, **kwargs))
         return variant
+
+    @staticmethod
+    def _lookup_variant_annotations():
+        return {}

hail_search/queries/snv_indel.py

@@ -2,7 +2,7 @@
 import hail as hl
 
 from hail_search.constants import GENOME_VERSION_GRCh38, SCREEN_KEY, PREFILTER_FREQ_CUTOFF, ALPHAMISSENSE_SORT, \
-    UTR_ANNOTATOR_KEY, EXTENDED_SPLICE_KEY, MOTIF_FEATURES_KEY, REGULATORY_FEATURES_KEY
+    UTR_ANNOTATOR_KEY, EXTENDED_SPLICE_KEY, MOTIF_FEATURES_KEY, REGULATORY_FEATURES_KEY, GENOME_VERSION_GRCh37
 from hail_search.queries.base import BaseHailTableQuery, PredictionPath
 from hail_search.queries.snv_indel_37 import SnvIndelHailTableQuery37
 
@@ -12,6 +12,7 @@
 class SnvIndelHailTableQuery(SnvIndelHailTableQuery37):
 
     GENOME_VERSION = GENOME_VERSION_GRCh38
+    LIFT_GENOME_VERSION = GENOME_VERSION_GRCh37
     PREDICTION_FIELDS_CONFIG = {
         **SnvIndelHailTableQuery37.PREDICTION_FIELDS_CONFIG,
         'fathmm': PredictionPath('dbnsfp', 'fathmm_MKL_coding_score'),
@@ -96,3 +97,7 @@ def _get_annotation_override_filters(self, ht, annotation_overrides):
                 )
 
         return annotation_filters
+
+    @staticmethod
+    def _lookup_variant_annotations():
+        return {'liftover_locus': lambda r: r.rg37_locus}

hail_search/queries/snv_indel_37.py

@@ -1,9 +1,10 @@
+from aiohttp.web import HTTPNotFound
 from collections import OrderedDict
 import hail as hl
 
 from hail_search.constants import CLINVAR_KEY, HGMD_KEY, HGMD_PATH_RANGES, \
     GNOMAD_GENOMES_FIELD, PREFILTER_FREQ_CUTOFF, PATH_FREQ_OVERRIDE_CUTOFF, PATHOGENICTY_HGMD_SORT_KEY, \
-    SPLICE_AI_FIELD, GENOME_VERSION_GRCh37
+    SPLICE_AI_FIELD, GENOME_VERSION_GRCh37, GENOME_VERSION_GRCh38
 from hail_search.queries.base import PredictionPath, QualityFilterFormat
 from hail_search.queries.mito import MitoHailTableQuery
 
@@ -12,6 +13,7 @@ class SnvIndelHailTableQuery37(MitoHailTableQuery):
 
     DATA_TYPE = 'SNV_INDEL'
     GENOME_VERSION = GENOME_VERSION_GRCh37
+    LIFT_GENOME_VERSION = GENOME_VERSION_GRCh38
 
     GENOTYPE_FIELDS = {f.lower(): f for f in ['DP', 'GQ', 'AB']}
     QUALITY_FILTER_FORMAT = {
@@ -133,3 +135,26 @@ def _get_annotation_override_filters(self, ht, annotation_overrides):
     @staticmethod
     def _stat_has_non_ref(s):
         return (s.het_samples > 0) | (s.hom_samples > 0)
+
+    @staticmethod
+    def _lookup_variant_annotations():
+        return {'liftover_locus': lambda r: r.rg38_locus}
+
+    @classmethod
+    def _get_lifted_table_path(cls, path):
+        return f'{cls._get_table_dir(path)}/{cls.LIFT_GENOME_VERSION}/{cls.DATA_TYPE}/{path}'
+
+    def _get_variant_project_data(self, variant_id, variant=None, **kwargs):
+        project_data = super()._get_variant_project_data(variant_id, **kwargs)
+        liftover_locus = variant.pop('liftover_locus')
+        if not liftover_locus:
+            return project_data
+        interval = hl.eval(hl.interval(liftover_locus, liftover_locus, includes_start=True, includes_end=True))
+        self._load_table_kwargs['_intervals'] = [interval]
+        self._get_table_path = self._get_lifted_table_path
+        try:
+            lift_project_data = super()._get_variant_project_data(variant_id, **kwargs)
+        except HTTPNotFound:
+            return project_data
+        project_data['familyGenotypes'].update(lift_project_data['familyGenotypes'])
+        return project_data.annotate(liftedFamilyGuids=sorted(lift_project_data['familyGenotypes'].keys()))

hail_search/test_utils.py

@@ -494,6 +494,7 @@
 
 VARIANT_LOOKUP_VARIANT = {
     **VARIANT1,
+    'liftedFamilyGuids': ['F000014_14'],
     'familyGenotypes': {
         VARIANT1['familyGuids'][0]: sorted([
             {k: v for k, v in g.items() if k != 'individualGuid'} for g in VARIANT1['genotypes'].values()
@@ -502,6 +503,10 @@
             'sampleId': 'NA20885', 'sampleType': 'WES', 'familyGuid': 'F000011_11',
             'numAlt': 2, 'dp': 6, 'gq': 16, 'ab': 1.0, 'filters': [],
         }],
+        'F000014_14': [{
+            'sampleId': 'NA21234', 'sampleType': 'WGS', 'familyGuid': 'F000014_14',
+            'numAlt': 1, 'dp': 27, 'gq': 87, 'ab': 0.531000018119812, 'filters': None,
+        }],
     }
 }
 for k in {'familyGuids', 'genotypes'}:

seqr/views/apis/variant_search_api.py

@@ -602,9 +602,11 @@ def _update_lookup_variant(variant, response):
             for genotype in variant['familyGenotypes'].pop(family_guid)
         })
 
-    for i, genotypes in enumerate(variant.pop('familyGenotypes').values()):
+    for i, (unmapped_family_guid, genotypes) in enumerate(variant.pop('familyGenotypes').items()):
         family_guid = f'F{i}_{variant["variantId"]}'
         variant['lookupFamilyGuids'].append(family_guid)
+        if unmapped_family_guid in variant.get('liftedFamilyGuids', []):
+            variant['liftedFamilyGuids'][variant['liftedFamilyGuids'].index(unmapped_family_guid)] = family_guid
         for j, genotype in enumerate(genotypes):
             individual_guid = f'I{j}_{family_guid}'
             individual = individual_summary_map[(genotype.pop('familyGuid'), genotype.pop('sampleId'))]

seqr/views/apis/variant_search_api_tests.py

@@ -804,12 +804,14 @@ def test_variant_lookup(self, mock_variant_lookup):
         expected_variant = {
             **VARIANT_LOOKUP_VARIANT,
             'familyGuids': [],
-            'lookupFamilyGuids': ['F0_1-10439-AC-A', 'F1_1-10439-AC-A'],
+            'lookupFamilyGuids': ['F0_1-10439-AC-A', 'F1_1-10439-AC-A', 'F2_1-10439-AC-A'],
+            'liftedFamilyGuids': ['F2_1-10439-AC-A'],
             'genotypes': {
                 'I0_F0_1-10439-AC-A': {'ab': 0.0, 'dp': 60, 'gq': 20, 'numAlt': 0, 'filters': [], 'sampleType': 'WES'},
                 'I1_F0_1-10439-AC-A': {'ab': 0.0, 'dp': 24, 'gq': 0, 'numAlt': 0, 'filters': [], 'sampleType': 'WES'},
                 'I2_F0_1-10439-AC-A': {'ab': 0.5, 'dp': 10, 'gq': 99, 'numAlt': 1, 'filters': [], 'sampleType': 'WES'},
                 'I0_F1_1-10439-AC-A': {'ab': 1.0, 'dp': 6, 'gq': 16, 'numAlt': 2, 'filters': [], 'sampleType': 'WES'},
+                'I0_F2_1-10439-AC-A': {'ab': 0.531000018119812, 'dp': 27, 'gq': 87, 'numAlt': 1, 'filters': None, 'sampleType': 'WGS'},
             },
         }
         del expected_variant['familyGenotypes']
@@ -830,6 +832,11 @@ def test_variant_lookup(self, mock_variant_lookup):
                     'features': [{'category': 'HP:0001626', 'label': '1 terms'}, {'category': 'Other', 'label': '1 terms'}],
                     'vlmContactEmail': 'seqr-test@gmail.com,test@broadinstitute.org',
                 },
+                'I0_F2_1-10439-AC-A': {
+                    'affected': 'A', 'familyGuid': 'F2_1-10439-AC-A', 'features': [],
+                    'individualGuid': 'I0_F2_1-10439-AC-A', 'sex': 'F',
+                    'vlmContactEmail': 'vlm@broadinstitute.org',
+                },
                 'I1_F0_1-10439-AC-A': {
                     'affected': 'N', 'familyGuid': 'F0_1-10439-AC-A', 'features': [],
                     'individualGuid': 'I1_F0_1-10439-AC-A', 'sex': 'M',
@@ -860,7 +867,7 @@ def test_variant_lookup(self, mock_variant_lookup):
 
         response_variant['variantId'] = '1-248367227-TC-T'
         response_variant['genomeVersion'] = '37'
-        self.login_collaborator()
+        self.login_manager()
         response = self.client.get(url.replace("38", "37"))
         self.assertEqual(response.status_code, 200)
 
@@ -869,9 +876,11 @@ def test_variant_lookup(self, mock_variant_lookup):
             ('I000005_hg00732', 'I1_F0_1-10439-AC-A', {'sampleId': 'HG00732', 'familyGuid': 'F000002_2'}),
             ('I000004_hg00731', 'I2_F0_1-10439-AC-A', {'sampleId': 'HG00731', 'familyGuid': 'F000002_2'}),
             ('I000015_na20885', 'I0_F1_1-10439-AC-A', {'sampleId': 'NA20885', 'familyGuid': 'F000011_11'}),
+            ('I000018_na21234', 'I0_F2_1-10439-AC-A', {'sampleId': 'NA21234', 'familyGuid': 'F000014_14'}),
         ]
         expected_variant.update({
-            'lookupFamilyGuids': ['F000002_2', 'F000011_11'],
+            'lookupFamilyGuids': ['F000002_2', 'F000011_11', 'F000014_14'],
+            'liftedFamilyGuids': ['F000014_14'],
             'genotypes': {
                 individual_guid: {**expected_variant['genotypes'][anon_individual_guid], **genotype}
                 for individual_guid, anon_individual_guid, genotype in individual_guid_map
@@ -881,32 +890,37 @@ def test_variant_lookup(self, mock_variant_lookup):
         })
         expected_body.update({
             **{k: {**EXPECTED_SEARCH_RESPONSE[k]} for k in {
-                'savedVariantsByGuid', 'variantTagsByGuid', 'variantNotesByGuid',
+                'mmeSubmissionsByGuid', 'variantTagsByGuid', 'variantNotesByGuid',
             }},
             **EXPECTED_TRANSCRIPTS_RESPONSE,
+            'omimIntervals': {},
+            'savedVariantsByGuid': {'SV0000002_1248367227_r0390_100': EXPECTED_SAVED_VARIANT},
             'variantFunctionalDataByGuid': {},
             'locusListsByGuid': EXPECTED_SEARCH_CONTEXT_RESPONSE['locusListsByGuid'],
             'projectsByGuid': {
                 p: {k: mock.ANY for k in PROJECT_TAG_TYPE_FIELDS}
-                for p in [PROJECT_GUID, 'R0003_test']
+                for p in [PROJECT_GUID, 'R0003_test', 'R0004_non_analyst_project']
             },
             'familiesByGuid': {
                 f: {k: mock.ANY for k in [*FAMILY_FIELDS, 'individualGuids']}
-                for f in ['F000002_2', 'F000011_11']
+                for f in ['F000002_2', 'F000011_11', 'F000014_14']
             },
             'individualsByGuid': {
                 i[0]: {k: mock.ANY for k in [*INDIVIDUAL_FIELDS, 'igvSampleGuids']}
-                for i in individual_guid_map
+                for i in individual_guid_map + [('I000019_na21987',)]
             },
         })
         expected_body['genesById']['ENSG00000227232'] = expected_pa_gene
-        del expected_body['savedVariantsByGuid']['SV0000001_2103343353_r0390_100']
+        expected_body['mmeSubmissionsByGuid']['MS000018_P0004517'] = expected_body['mmeSubmissionsByGuid'].pop('MS000001_na19675')
+        expected_body['savedVariantsByGuid']['SV0000006_1248367227_r0004_non'] = mock.ANY
+        expected_body['variantTagsByGuid']['VT1726970_2103343353_r0004_tes'] = EXPECTED_TAG
+        expected_body['variantTagsByGuid']['VT1726961_2103343353_r0005_tes'] = EXPECTED_TAG
         for k in ['VT1708633_2103343353_r0390_100', 'VT1726961_2103343353_r0390_100']:
             del expected_body['variantTagsByGuid'][k]
 
         self.assertDictEqual(response.json(), expected_body)
         mock_variant_lookup.assert_called_with(
-            self.collaborator_user, ('1', 10439, 'AC', 'A'), genome_version='37',
+            self.manager_user, ('1', 10439, 'AC', 'A'), genome_version='37',
         )
 
     @mock.patch('seqr.views.apis.variant_search_api.sv_variant_lookup')

ui/pages/SummaryData/components/VariantLookup.jsx

@@ -1,7 +1,7 @@
 import React from 'react'
 import { connect } from 'react-redux'
 import PropTypes from 'prop-types'
-import { Grid, Header } from 'semantic-ui-react'
+import { Grid, Header, Label } from 'semantic-ui-react'
 
 import { RECEIVE_DATA } from 'redux/utils/reducerUtils'
 import { QueryParamsEditor } from 'shared/components/QueryParamEditor'
@@ -14,7 +14,7 @@ import FamilyReads from 'shared/components/panel/family/FamilyReads'
 import FamilyVariantTags from 'shared/components/panel/variants/FamilyVariantTags'
 import Variants, { Variant, StyledVariantRow } from 'shared/components/panel/variants/Variants'
 import { FamilyVariantIndividuals } from 'shared/components/panel/variants/VariantIndividuals'
-import { GENOME_VERSION_FIELD } from 'shared/utils/constants'
+import { GENOME_VERSION_FIELD, GENOME_VERSION_37, GENOME_VERSION_38 } from 'shared/utils/constants'
 import { sendVlmContactEmail } from '../reducers'
 import { getVlmDefaultContactEmails, getVlmFamiliesByContactEmail } from '../selectors'
 
@@ -43,10 +43,19 @@ const mapContactDispatchToProps = {
 
 const ContactButton = connect(null, mapContactDispatchToProps)(SendEmailButton)
 
-const LookupFamilyLayout = ({ topContent, bottomContent, children, ...buttonProps }) => (
+const liftoverGenomeVersion = genomeVersion => (
+  genomeVersion === GENOME_VERSION_37 ? GENOME_VERSION_38 : GENOME_VERSION_37
+)
+
+const LookupFamilyLayout = ({ topContent, bottomContent, hasLiftover, genomeVersion, children, ...buttonProps }) => (
   <StyledVariantRow>
     {topContent}
     <Grid.Column width={4}>
+      <Label
+        content={`GRCh${hasLiftover ? liftoverGenomeVersion(genomeVersion) : genomeVersion}`}
+        basic
+        color={hasLiftover ? 'orange' : 'green'}
+      />
       <ContactButton {...buttonProps} />
     </Grid.Column>
     <Grid.Column width={12}>
@@ -60,6 +69,8 @@ LookupFamilyLayout.propTypes = {
   topContent: PropTypes.node,
   bottomContent: PropTypes.node,
   children: PropTypes.node,
+  hasLiftover: PropTypes.bool,
+  genomeVersion: PropTypes.string,
 }
 
 const InternalFamily = ({ familyGuid, variant, reads, showReads }) => (
@@ -70,6 +81,8 @@ const InternalFamily = ({ familyGuid, variant, reads, showReads }) => (
       </Grid.Column>
     )}
     bottomContent={<Grid.Column width={16}>{reads}</Grid.Column>}
+    hasLiftover={variant.liftedFamilyGuids?.includes(familyGuid)}
+    genomeVersion={variant.genomeVersion}
   >
     <FamilyVariantIndividuals familyGuid={familyGuid} variant={variant} />
     {showReads}
@@ -96,6 +109,8 @@ const BaseLookupVariant = ({ variant, familiesByContactEmail, vlmDefaultContactE
           key={contactEmail}
           defaultEmail={vlmDefaultContactEmails[contactEmail]}
           modalId={contactEmail}
+          hasLiftover={(variant.liftedFamilyGuids || []).some(familyGuid => families.includes(familyGuid))}
+          genomeVersion={variant.genomeVersion}
         >
           <Grid stackable divided="vertically">
             {families.map(familyGuid => (