Fix problem with the gene counts in merge_results

jfnavarro · Feb 12, 2021 · 7676579 · 7676579
1 parent 56837d8
commit 7676579
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 29 deletions.
diff --git a/hlapipeline/variants.py b/hlapipeline/variants.py
@@ -24,38 +24,12 @@ def __init__(self):
         self.dbsnp = None
         self.gnomad = None
         self.cosmic = None
+        self.gene = None
 
     @property
     def key(self):
         return "{}:{} {}>{}".format(self.chrom, self.start, self.ref, self.alt)
 
-    @property
-    def ensGene(self):
-        gene = None
-        for effect in self.epitopes:
-            if 'ensGene' in effect.func:
-                gene = effect.func.split("_")[1]
-                break
-        return gene
-
-    @property
-    def knownGene(self):
-        gene = None
-        for effect in self.epitopes:
-            if 'knownGene' in effect.func:
-                gene = effect.func.split("_")[1]
-                break
-        return gene
-
-    @property
-    def refGene(self):
-        gene = None
-        for effect in self.epitopes:
-            if 'refGene' in effect:
-                gene = effect.func.split("_")[1]
-                break
-        return gene
-
     def __str__(self):
         return '{}:{} {}>{} {} {}'.format(self.chrom, self.start, self.ref, self.alt, self.type, self.status)
 
@@ -141,6 +115,8 @@ def filter_variants_rna(file, tumor_coverage, tumor_var_depth,
         gnomad_AF = record.INFO['AF'][0] if record.INFO['AF'] != [] else 'NA'
         cosmic70 = ';'.join(record.INFO['cosmic70']).split(":")[1].split("-")[0] if record.INFO[
                                                                                         'cosmic70'] != [] else 'NA'
+        gene = record.INFO['Gene.knownGene'][0] if record.INFO['Gene.knownGene'] != [] else None
+
         if has_func_ens or has_func_known or has_func_ref:
             called = {x.sample: x.data for x in record.calls if x.called}
             filtered = dict()
@@ -177,6 +153,7 @@ def filter_variants_rna(file, tumor_coverage, tumor_var_depth,
             variant.gnomad = gnomad_AF
             variant.cosmic = cosmic70
             variant.type = 'rna'
+            variant.gene = gene
             variants.append(variant)
 
     return variants
@@ -218,6 +195,7 @@ def filter_variants_dna(file, normal_coverage, tumor_coverage, tumor_var_depth,
         gnomad_AF = record.INFO['AF'][0] if record.INFO['AF'] != [] else 'NA'
         cosmic70 = ';'.join(record.INFO['cosmic70']).split(":")[1].split("-")[0] if record.INFO[
                                                                                         'cosmic70'] != [] else 'NA'
+        gene = record.INFO['Gene.knownGene'][0] if record.INFO['Gene.knownGene'] != [] else None
 
         if has_func_ens or has_func_known or has_func_ref:
             called = {x.sample: x.data for x in record.calls if x.called}
@@ -351,6 +329,7 @@ def filter_variants_dna(file, normal_coverage, tumor_coverage, tumor_var_depth,
             variant.gnomad = gnomad_AF
             variant.cosmic = cosmic70
             variant.type = 'dna'
+            variant.gene = gene
             variants.append(variant)
 
     return variants
diff --git a/merge_results.py b/merge_results.py
@@ -154,8 +154,7 @@ def main(dna_variants,
         dbsnp = value[0][0].dbsnp
         gnomad = value[0][0].gnomad
         cosmic = value[0][0].cosmic
-        # Get the gene information of the gene (we want to make sure we use the gene name annotation)
-        gene = value[0][0].knownGene
+        gene = value[0][0].gene
 
         # Create a dictionary of epitopes so to keep unique ones (different mut peptide)
         epitopes_dict = defaultdict(list)