allele frequencies modifications

EGA-archive · Oct 17, 2024 · 50aaf62 · 50aaf62
1 parent 9b2d91f
commit 50aaf62
Show file tree

Hide file tree

Showing 2 changed files with 52 additions and 37 deletions.
diff --git a/conf/conf.py b/conf/conf.py
@@ -5,8 +5,8 @@
 #### VCF Conversion config parameters ####
 allele_frequency=1 # introduce float number, leave 1 if you want to convert all the variants
 reference_genome='GRCh38' # Choose one between NCBI36, GRCh37, GRCh38
-datasetId='CINECA_synthetic_cohort_EUROPE_UK1'
-case_level_data=True
+datasetId='COVID_pop11_fin_2'
+case_level_data=False
 
 ### MongoDB parameters ###
 database_host = 'mongo'

diff --git a/genomicVariations_vcf.py b/genomicVariations_vcf.py
@@ -86,7 +86,7 @@ def num_rows_in_vcf_files():
             total_lines += sum(1 for line in f if not line.startswith('#'))
     return total_lines
 
-num_rows = num_rows_in_vcf_files()
+num_rows = 1000000000
 
 def generate(dict_properties):
     total_dict =[]
@@ -105,13 +105,7 @@ def generate(dict_properties):
             #print(v)
             dict_to_xls={}
             vstringed = str(v)
-            for population_splitted in pipeline['frequencyInPopulations|frequencies|population']:
-                splitword=population_splitted['fullname']+'='
-                v_splitted = vstringed.split(splitword)
-                if len(v_splitted) < 2:
-                    continue
-                else:
-                    population = population_splitted['shortname']
+            population='COVID_pop11_fin_2'
 
             if conf.case_level_data == True:
                 clinicalRelevanceword=pipeline['caseLevelData|clinicalInterpretations|clinicalRelevance']+'='
@@ -159,36 +153,53 @@ def generate(dict_properties):
                 except Exception:
                     pass
             try:
-                v_resplitted = v_splitted[1].split(';')
-                allele_frequency = v_resplitted[0]
-                try:
-                    allele_frequency = float(allele_frequency)
-                except Exception:
-                    allele_frequency = allele_frequency.split(',')
-                    allele_frequency=float(allele_frequency[0])
-                if allele_frequency != '' and population != '':
-                    dict_to_xls['frequencyInPopulations|sourceReference']='gnomad.broadinstitute.org/'
-                    dict_to_xls['frequencyInPopulations|source']='The Genome Aggregation Database (gnomAD)'
-                    dict_to_xls['frequencyInPopulations|frequencies|population']=population
-                    dict_to_xls['frequencyInPopulations|frequencies|alleleFrequency']=allele_frequency
-            except Exception:
-                try:
-                    allele_frequency=v.INFO.get('AF')
-                    if isinstance(allele_frequency, tuple):
-                        allele_frequency=list(allele_frequency)
-                        allele_frequency[0]
-                    else:
-                        allele_frequency = float(v.INFO.get('AF'))
-                    allele_number = float(v.INFO.get('AN'))
+                allele_frequency=v.INFO.get('AF')
+                if isinstance(allele_frequency, tuple):
+                    allele_frequency=list(allele_frequency)
+                    allele_frequency[0]
+                else:
+                    allele_frequency = float(v.INFO.get('AF'))
+                if allele_frequency == 0.0:
+                    continue
+                allele_number=v.INFO.get('AN')
+                if allele_number == None:
+                    pass
+                elif isinstance(allele_number, tuple):
+                    allele_number=list(allele_number)
+                    allele_number[0]
+                else:
+                    allele_number = float(v.INFO.get('AC'))
+                allele_count=v.INFO.get('AC')
+                if allele_count == None:
+                    pass
+                elif isinstance(allele_count, tuple):
+                    allele_count=list(allele_count)
+                    allele_count[0]
+                else:
                     allele_count = float(v.INFO.get('AC'))
+                ac_hom=v.INFO.get('AC_Hom')
+                if ac_hom == None:
+                    pass
+                elif isinstance(ac_hom, tuple):
+                    ac_hom=list(ac_hom)
+                    ac_hom[0]
+                else:
                     ac_hom = float(v.INFO.get('AC_Hom'))
-                    ac_het= float(v.INFO.get('AC_Het'))
-                    dict_to_xls['frequencyInPopulations|sourceReference']=pipeline["frequencyInPopulations|sourceReference"]
-                    dict_to_xls['frequencyInPopulations|source']=pipeline["frequencyInPopulations|source"]
-                    dict_to_xls['frequencyInPopulations|frequencies|population']=pipeline["frequencyInPopulations|frequencies|population"][0]["fullname"]
-                    dict_to_xls['frequencyInPopulations|frequencies|alleleFrequency']=allele_frequency
-                except Exception:
+                ac_het=v.INFO.get('AC_Het')
+                if ac_het == None:
                     pass
+                elif isinstance(ac_het, tuple):
+                    ac_het=list(ac_het)
+                    ac_het[0]
+                else:
+                    ac_het = float(v.INFO.get('AC_Het'))
+
+                dict_to_xls['frequencyInPopulations|sourceReference']=pipeline["frequencyInPopulations|sourceReference"]
+                dict_to_xls['frequencyInPopulations|source']=pipeline["frequencyInPopulations|source"]
+                dict_to_xls['frequencyInPopulations|frequencies|population']=population
+                dict_to_xls['frequencyInPopulations|frequencies|alleleFrequency']=allele_frequency
+            except Exception as e:
+                continue
             try:
                 if v.INFO.get('VT') == 'SV': continue
             except Exception:
@@ -222,6 +233,7 @@ def generate(dict_properties):
             except Exception:
                 dict_to_xls['variation|variantType']='UNKNOWN'
             #print(v.INFO.get('ANN'))
+            '''
             if v.INFO.get('ANN') is not None:
                 annot = v.INFO.get('ANN')
                 transcripts = annot.split(',')
@@ -354,6 +366,9 @@ def generate(dict_properties):
                         dict_to_xls['molecularAttributes|aminoacidChanges'] = annotations[10]
                     dict_to_xls['molecularAttributes|geneIds'] = annotations[4]
 
+
+            '''
+
 
 
             zigosity={}