refactor genotyping scripts

workflow/scripts/genotyping/combine_all_alleles.py

@@ -0,0 +1,28 @@
+import os
+import sys
+
+def main():
+    infiles = sys.argv[1].split(' ')
+    alleles = {}
+    for mhc in infiles:
+        fh_in = open(mhc, 'r')
+        # skip header
+        for line in fh_in:
+            cols = line.rstrip().split('\t')
+            source = cols[0]
+            mhc = cols[1]
+
+            if mhc not in alleles:
+                alleles[mhc] = source
+            else:
+                alleles[mhc] += ',' + source
+        fh_in.close()
+
+    outfile = sys.argv[2]
+    fh_out = open(outfile, 'w')
+    for allele in alleles:
+        fh_out.write(f'{alleles[allele]}\t{allele}\n')
+    fh_out.close()
+
+
+main()

workflow/scripts/genotyping/merge_predicted_mhcI.py

@@ -0,0 +1,32 @@
+import os
+import sys
+import re
+from pathlib import Path
+
+def main():
+    files = sys.argv[1]
+    alleles = {}
+    for file in files.rstrip().split(' '):
+        filestem = Path(file).stem
+        se = re.search(r'^(.+)_(RNA|DNA)_(SE|PE)', filestem)
+        group = se.group(1)
+
+        fh = open(file, 'r')
+        for line in fh:
+            al = line.rstrip()
+            if al not in alleles:
+                alleles[al] = []
+
+            alleles[al].append(group)
+
+
+    out = open(sys.argv[2], 'w')
+    for al in dict(sorted(alleles.items())):
+        for i,v in enumerate(alleles[al]):
+            if i == 0:
+                out.write(v)
+            else:
+                out.write(f',{v}')
+        out.write(f'\t{al}\n')
+
+main()