Dev

clickhouse_search/backend/functions.py

@@ -95,6 +95,13 @@ def process_rhs(self, compiler, connection):
         return rhs_params[0], []
 
 
+@NestedField.register_lookup
+@ArrayField.register_lookup
+class ArrayAll(ArrayExists):
+    lookup_name = "array_all"
+    function = "arrayAll"
+
+
 class ArrayFilter(lookups.Transform):
     def __init__(self, *args, conditions=None, **kwargs):
         super().__init__(*args, **kwargs)
@@ -113,6 +120,20 @@ class ArrayNotEmptyTransform(lookups.Transform):
     output_field = BooleanField()
 
 
+@ArrayField.register_lookup
+class BitmapHasAny(ArrayLookup):
+    lookup_name = "bitmap_has_any"
+    function = "bitmapHasAny"
+
+    def process_lhs(self, compiler, connection):
+        lhs, lhs_params = super().process_lhs(compiler, connection)
+        return f'bitmapBuild({lhs})', lhs_params
+
+    def process_rhs(self, compiler, connection):
+        rhs, rhs_params = super().process_rhs(compiler, connection)
+        return f'bitmapBuild{rhs.split("::")[0]}', rhs_params
+
+
 class DictGet(Func):
     function = 'dictGet'
     template = '%(function)s("%(dict_name)s", (%(fields)s), %(expressions)s)'

clickhouse_search/fixtures/clickhouse_saved_variants.json

@@ -66,6 +66,8 @@
       "sample_type": "WGS",
       "xpos": 1248367227,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [6, 48],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -118,6 +120,8 @@
       "sample_type": "WES",
       "xpos": 2103343353,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [48, 45],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -196,6 +200,8 @@
       "sample_type": "WGS",
       "xpos": 1248367227,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [

clickhouse_search/fixtures/clickhouse_search.json

@@ -274,6 +274,8 @@
       "sample_type": "WES",
       "xpos": 1000010439,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -292,6 +294,8 @@
       "sample_type": "WES",
       "xpos": 1038724419,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [60, 72],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -310,6 +314,8 @@
       "sample_type": "WES",
       "xpos": 1091502721,
       "is_gnomad_gt_5_percent": true,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61, 60],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -328,6 +334,70 @@
       "sample_type": "WES",
       "xpos": 1091511686,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61],
+      "filters": [
+        "VQSRTrancheSNP99.95to100.00"
+      ],
+      "sign": 1,
+      "calls": [
+        ["HG00733", 0, 0, 0, 45],
+        ["HG00732", 0, 0, 0, 24],
+        ["HG00731", 1, 58, 0.17241, 29]
+      ]
+    }
+}, {
+    "model": "clickhouse_search.entriessnvindel",
+    "pk": 52,
+    "fields": {
+      "key": 2,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "sample_type": "WES",
+      "xpos": 1038724419,
+      "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [60, 72],
+      "filters": [],
+      "sign": 1,
+      "calls": [
+        ["HG00733", 0, 40, 0, 33],
+        ["HG00732", 1, 99, 0.625, 32],
+        ["HG00731", 2, 99, 1, 36]
+      ]
+    }
+}, {
+    "model": "clickhouse_search.entriessnvindel",
+    "pk": 53,
+    "fields": {
+      "key": 3,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "sample_type": "WES",
+      "xpos": 1091502721,
+      "is_gnomad_gt_5_percent": true,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61, 60],
+      "filters": [],
+      "sign": 1,
+      "calls": [
+        ["HG00733", 1, 99, 0.40741, 27],
+        ["HG00732", 0, 99, 0.45946, 37],
+        ["HG00731", 1, 99, 1, 40]
+      ]
+    }
+}, {
+    "model": "clickhouse_search.entriessnvindel",
+    "pk": 54,
+    "fields": {
+      "key": 4,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "sample_type": "WES",
+      "xpos": 1091511686,
+      "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61],
       "filters": [
         "VQSRTrancheSNP99.95to100.00"
       ],
@@ -348,10 +418,11 @@
       "sample_type": "WES",
       "xpos": 1091502721,
       "is_gnomad_gt_5_percent": true,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61, 60],
       "filters": [],
       "sign": 1,
       "calls": [
-        ["NA20872", 1, 99, 0.34285714285714286, 35],
         ["NA20870", 1, 99, 0.6785714285714286, 28]
       ]
     }
@@ -365,6 +436,8 @@
       "sample_type": "WES",
       "xpos": 1000010439,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -381,6 +454,8 @@
       "sample_type": "WES",
       "xpos": 1038724419,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [60, 72],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -397,6 +472,8 @@
       "sample_type": "WES",
       "xpos": 1000010146,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -413,6 +490,8 @@
       "sample_type": "WGS",
       "xpos": 1000010439,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -429,6 +508,8 @@
       "sample_type": "WGS",
       "xpos": 1038724419,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [60, 72],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -445,6 +526,8 @@
       "sample_type": "WGS",
       "xpos": 1000010146,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -461,6 +544,8 @@
       "sample_type": "WGS",
       "xpos": 1000010439,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -479,6 +564,8 @@
       "sample_type": "WGS",
       "xpos": 1038724419,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [60, 72],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -497,6 +584,8 @@
       "sample_type": "WGS",
       "xpos": 1091502721,
       "is_gnomad_gt_5_percent": true,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61, 60],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -515,6 +604,8 @@
       "sample_type": "WGS",
       "xpos": 1091511686,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [61],
       "filters": [
         "VQSRTrancheSNP99.95to100.00"
       ],
@@ -535,6 +626,8 @@
       "sample_type": "WGS",
       "xpos": 1009310123,
       "is_gnomad_gt_5_percent": false,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [71],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -581,6 +674,8 @@
       "sample_type": "WES",
       "xpos": 7143270172,
       "is_gnomad_gt_5_percent": true,
+      "is_annotated_in_any_gene": true,
+      "geneId_ids": [73, 74],
       "filters": [
         "VQSRTrancheSNP99.90to99.95"
       ],
@@ -600,6 +695,8 @@
       "sample_type": "WGS",
       "xpos": 1000010439,
       "is_gnomad_gt_5_percent": true,
+      "is_annotated_in_any_gene": false,
+      "geneId_ids": [],
       "filters": [],
       "sign": 1,
       "calls": [
@@ -1116,6 +1213,66 @@
         ["HG00733", 1, 3, 29, false, 38721781, 38734440, 7, ["ENSG00000275023", "ENSG00000277258", "ENSG00000277972"], false, true, false]
       ]
     }
+}, {
+    "model": "clickhouse_search.entriesgcnv",
+    "pk": 48,
+    "fields": {
+      "key": 16,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "xpos": 14022417556,
+      "filters": [],
+      "sign": 1,
+      "calls": [
+        ["HG00731", 1, 3, 38, false, 22438910, 22469796, 0, [], false, true, false],
+        ["HG00733", null, null, null, null, null, null, null, [], null, null, null]
+      ]
+    }
+}, {
+    "model": "clickhouse_search.entriesgcnv",
+    "pk": 49,
+    "fields": {
+      "key": 17,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "xpos": 16029802672,
+      "filters": [],
+      "sign": 1,
+      "calls": [
+        ["HG00731", 1, 3, 29, false, 29809156, 29815990, 8, ["ENSG00000103495", "ENSG00000167371", "ENSG00000280893"], false, true, false],
+        ["HG00733", 1, 3, 37, false, 29809156, 29815990, 8, ["ENSG00000103495", "ENSG00000167371", "ENSG00000280893"], false, true, false]
+      ]
+    }
+}, {
+    "model": "clickhouse_search.entriesgcnv",
+    "pk": 40,
+    "fields": {
+      "key": 18,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "xpos": 17038717327,
+      "filters": [],
+      "sign": 1,
+      "calls": [
+        ["HG00731", 2, 4, 13, true, 38717327, 38719636, 3, ["ENSG00000275023"], true, false, false],
+        ["HG00733", null, null, null, null, null, null, null, [], null, null, null]
+      ]
+    }
+}, {
+    "model": "clickhouse_search.entriesgcnv",
+    "pk": 41,
+    "fields": {
+      "key": 19,
+      "project_guid": "R0001_1kg",
+      "family_guid": "F000002_2_x",
+      "xpos": 17038721781,
+      "filters": [],
+      "sign": 1,
+      "calls": [
+        ["HG00731", 1, 3, 28, false, 38721781, 38735703, 7, ["ENSG00000275023", "ENSG00000277258", "ENSG00000277972"], false, true, false],
+        ["HG00733", 1, 3, 29, false, 38721781, 38734440, 7, ["ENSG00000275023", "ENSG00000277258", "ENSG00000277972"], false, true, false]
+      ]
+    }
 }, {
     "model": "clickhouse_search.entriesgcnv",
     "pk": 32,
@@ -1339,5 +1496,35 @@
         "gencode_gene_type": "protein_coding",
         "gencode_release": 27
     }
+},
+{
+    "model": "reference_data.geneinfo",
+    "pk": 73,
+    "fields": {
+        "gene_id": "ENSG00000271079",
+        "gene_symbol": "CTAGE15",
+        "chrom_grch38": "7",
+        "start_grch38": 143571801,
+        "end_grch38": 143574387,
+        "strand_grch38": "+",
+        "coding_region_size_grch38": 0,
+        "gencode_gene_type": "protein_coding",
+        "gencode_release": 27
+    }
+},
+{
+    "model": "reference_data.geneinfo",
+    "pk": 74,
+    "fields": {
+        "gene_id": "ENSG00000176227",
+        "gene_symbol": "CTAGE6",
+        "chrom_grch38": "7",
+        "start_grch38": 91500851,
+        "end_grch38": 91525764,
+        "strand_grch38": "+",
+        "coding_region_size_grch38": 0,
+        "gencode_gene_type": "protein_coding",
+        "gencode_release": 27
+    }
 }
 ]