Merge pull request #5 from eastgenomics/fix_things

simplify de novo filtering (#5)
eastgenomics · Jun 28, 2024 · b438e23 · b438e23
2 parents 599d2a3 + 5a8f0ae
commit b438e23
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 91 deletions.
diff --git a/resources/home/dnanexus/make_workbook.py b/resources/home/dnanexus/make_workbook.py
@@ -622,20 +622,6 @@ def create_gel_tiering_variant_page(self):
             else:
                 summary_sheet[key] = 0
 
-    def get_de_novo_threshold(self, variant):
-        '''
-        The de novo quality score thresholds are different for indels or SNVs
-        This function works out if a variant is an indel based on having either
-        a ref or an alt sequence > 1
-        '''
-        if (len(variant["variantCoordinates"]["reference"]) > 1 or
-                len(variant["variantCoordinates"]["alternate"]) > 1):
-            threshold = self.config['denovo_quality_scores']['indel']
-        else:
-            threshold = self.config['denovo_quality_scores']['snv']
-
-        return threshold
-
     def create_additional_analysis_page(self):
         '''
         Get Tier3/Null SNVs for Exomiser/deNovo analysis
@@ -700,35 +686,32 @@ def create_additional_analysis_page(self):
             variant_list.append(var_dict)
 
         # Get variants with high de novo quality score (these are either SNVs
-        # or indels)
+        # or indels). These variants only appear in the JSON if the quality
+        # score is above the threshold
         for snv in self.wgs_data["interpretedGenomes"][
                 self.gel_index
             ]["interpretedGenomeData"]["variants"]:
             for event in snv["reportEvents"]:
-                if event['deNovoQualityScore'] is not None:
-                    if (
-                        event['deNovoQualityScore'] >
-                        self.get_de_novo_threshold(snv)
-                        ):
-                        event_index = snv["reportEvents"].index(event)
-                        var_dict = VariantUtils.get_snv_info(
+                if event['segregationPattern'] == 'deNovo':
+                    event_index = snv["reportEvents"].index(event)
+                    var_dict = VariantUtils.get_snv_info(
+                        snv,
+                        self.proband,
+                        event_index,
+                        self.column_list,
+                        self.mother,
+                        self.father,
+                        self.proband_sex
+                    )
+                    var_dict["Priority"] = "De novo"
+                    var_dict["Inheritance"] = "De novo"
+                    var_dict["HGVSc"], var_dict["HGVSp"] = (
+                        VariantNomenclature.get_hgvs_gel(
                             snv,
-                            self.proband,
-                            event_index,
-                            self.column_list,
-                            self.mother,
-                            self.father,
-                            self.proband_sex
-                        )
-                        var_dict["Priority"] = "De novo"
-                        var_dict["Inheritance"] = "De novo"
-                        var_dict["HGVSc"], var_dict["HGVSp"] = (
-                            VariantNomenclature.get_hgvs_gel(
-                                snv,
-                                self.mane,
-                                self.refseq_tsv)
-                        )
-                        variant_list.append(var_dict)
+                            self.mane,
+                            self.refseq_tsv)
+                    )
+                    variant_list.append(var_dict)
 
         ex_df = pd.DataFrame(variant_list)
         ex_df = ex_df.drop_duplicates()

diff --git a/resources/home/dnanexus/tests/test_make_workbook.py b/resources/home/dnanexus/tests/test_make_workbook.py
@@ -253,56 +253,3 @@ def test_get_ensp(self):
         assert VariantNomenclature.get_ensp(
             refseq_tsv, "ENST0000033"
         ) == "ENSP0000044"
-
-
-class TestDeNovoThresholds():
-    '''
-    Tests for function that gets the de novo quality score threshold based on
-    the ref and alt of the variant. The threshold should be different for
-    indels and SNVs, so this test assumes that a ref or alt longer than one
-    base is an indel.
-    '''
-    variants = [
-        {
-            "variantCoordinates": {
-                "reference": "GT",
-                "alternate": "G"
-                },
-        },
-        {
-            "variantCoordinates": {
-                "reference": "C",
-                "alternate": "G"
-                },
-        },
-        {
-            "variantCoordinates": {
-                "reference": "A",
-                "alternate": "AC"
-                },
-        },
-    ]
-    config = {
-        "denovo_quality_scores": {
-            "indel": 0.02,
-            "snv": 0.0013
-        }
-    }
-
-    def test_deletion_de_novo_threshold_correct(self):
-        '''
-        Test that get_de_novo_threshold gets correct threshold for an deletion
-        '''
-        assert excel.get_de_novo_threshold(self, self.variants[0]) == 0.02
-
-    def test_snv_de_novo_threshold_correct(self):
-        '''
-        Test that get_de_novo_threshold gets correct threshold for an SNV
-        '''
-        assert excel.get_de_novo_threshold(self, self.variants[1]) == 0.0013
-
-    def test_insertion_de_novo_threshold_correct(self):
-        '''
-        Test that get_de_novo_threshold gets correct threshold for an insertion
-        '''
-        assert excel.get_de_novo_threshold(self, self.variants[2]) == 0.02