added ClinVar_to_SAPA_parser and some fixes to the main

Author: Tobias Juhre

ClinVar_to_SAPA_parser.py

@@ -0,0 +1,45 @@
+import csv
+import sys
+
+print "converting file to SAPA format"
+
+counter = 0
+clinvar = open('converted_hg38_pathogenic.csv', 'w')
+clinvar.write("Chr,Pos,Ref,Alt,Clinical significance,Gene\n")
+with open("clinvar_pathogenic_snp.txt") as csvfile:
+    variant_lines = csv.reader(csvfile, delimiter='\t', quotechar='"')
+    # skip header if there
+    has_header = csv.Sniffer().has_header(csvfile.read(100))
+    csvfile.seek(0)  # rewind
+    incsv = csv.reader(csvfile)
+    if has_header:
+        header = next(variant_lines)
+        print header
+    for variant_line in variant_lines:
+        export_string = ""
+        ref_alt = variant_line[0].split(">")
+        if len(ref_alt) == 2:
+            ref = ref_alt[0][-1]
+            alt = ref_alt[1][0]
+            chr = variant_line[6]
+            pos = variant_line[7]
+            clin_sig = variant_line[4].split("(")
+            buildver = variant_line[8]
+            gene = variant_line[1]
+
+            if buildver == "GRCh38" and chr and pos:  # and clin_sig == "Pathogenic"
+                counter += 1
+                print chr
+                print pos
+                print ref
+                print alt
+                print clin_sig[0]
+                print gene
+                # print buildver
+                export_string += chr+","+pos+","+ref+","+alt+","+clin_sig[0]+","+gene+"\n"
+                clinvar.write(export_string)
+
+        if counter == 100:
+            break
+clinvar.close()
+print "END"

annovarParser_ljb26_all.py

@@ -149,9 +149,3 @@ def export_tab(self):
                          self.__phyloP46way_placental,
                          self.__phyloP100way_vertebrate,
                          self.__SiPhy_29way_logOdds)
-
-#                         self.__g2014oct_all,
-#                         self.__g2014oct_afr,
-#                         self.__g2014oct_eas,
-#                         self.__g2014oct_eur,
-0Suchvorgang...

main.py

@@ -140,7 +140,7 @@
         variant_lines = csv.reader(csvfile, delimiter=',', quotechar='"')
 
     # skip header if there
-    has_header = csv.Sniffer().has_header(csvfile.read(100))
+    has_header = csv.Sniffer().has_header(csvfile.read(25)) # 100
     csvfile.seek(0)  # rewind
     incsv = csv.reader(csvfile)
     if has_header:
@@ -163,14 +163,14 @@
                             variant_line[4], context, consequences, variant_line[7], variant_line[8], variant_line[9],
                             int(variant_line[10]), variant_line[11], variant_line[12], variant_line[13],
                             variant_line[14], variant_line[15]))
-        elif len(variant_line) < 16:
+        elif 16 > len(variant_line) > 4:
             success_count += 1
             # context = variant_line[5].split(",")
             # consequences = variant_line[6].split(",")
             snps.append(SNP(l_count, variant_line[0], int(variant_line[1]), variant_line[2], variant_line[3],
-                            ".", ".", ".", ".", ".", ".", ".", ".", ".", ".", ".", "."))
+                            variant_line[4], variant_line[5], ".", ".", ".", ".", ".", ".", ".", ".", ".", "."))
         else:
-            print "INVALID DATA (length < 16) in Line {}".format(l_count)
+            print "INVALID DATA (16 >= length > 4) in Line {}".format(l_count)
             print variant_line
             fail_count += 1
         l_count += 1
@@ -230,7 +230,7 @@
 databases = ["-buildver " + buildversion + " -downdb -webfrom annovar refGene " + buildversion,
              "-buildver " + buildversion + " -downdb cytoBand " + buildversion,
              "-buildver " + buildversion + " -downdb -webfrom annovar esp6500siv2_all " + buildversion,
-             "-buildver " + buildversion + " -downdb -webfrom annovar avsnp147 " + buildversion,  # snp138
+             "-buildver " + buildversion + " -downdb -webfrom annovar avsnp147 " + buildversion,  # avsnp147
              "-buildver " + buildversion + " -downdb -webfrom annovar dbnsfp30a " + buildversion
              ]
 
@@ -294,7 +294,7 @@
 else:
     params = "amplicon_variants_tab.csv " + buildversion + " -buildver " + buildversion + " -out myanno -remove -protocol " \
                                                                "refGene,cytoBand,esp6500siv2_all,avsnp147,dbnsfp30a " \
-                                                               "-operation g,r,f,f,f -nastring . "  #snp138
+                                                               "-operation g,r,f,f,f -nastring . "  #avsnp147
 if args.quiet:
     FNULL = open(os.devnull, 'w')
     p = subprocess.Popen([annovar_pl + params], shell=True, stdout=FNULL, stderr=subprocess.STDOUT)