Demultiplex in gzipped format.

fastq-demultiplexing/Python/bin/qaf-demux

@@ -12,6 +12,7 @@ import argparse
 import fileinput
 import os
 from itertools import islice
+from gzip import open as gzopen
 # from qaf_demux.libqafdemux import parse_fastq
 
 OPD = os.path.dirname
@@ -70,15 +71,16 @@ def main():
             bc_len = len(barcode)
         else:
             assert(len(barcode) == bc_len), f"len({barcode}) != {bc_len}"
-        outfiles[barcode] = open(OPJ(out_dir, f"{barcode}.fastq"), "w")
+        outfiles[barcode] = gzopen(OPJ(out_dir, f"{barcode}.fastq.gz"), "wb")
     if not outfiles:
         print("No barcode given.")
         exit(0)
-    outfiles["undetermined"] = open(OPJ(
-        out_dir, f"undetermined.fastq"), "w")
+    outfiles["undetermined"] = gzopen(OPJ(
+        out_dir, f"undetermined.fastq.gz"), "wb")
     bc_start = args.barcode_start - 1
 
     def bc_prob_profile(seq, qual):
+        # if bytes instead of string
         # return [(letter, 10 ** ((33 - quality) / 10))
         return [(letter, 10 ** ((33 - ord(quality)) / 10))
                 for letter, quality
@@ -125,7 +127,7 @@ def main():
                      best_prob,
                      best_diff) = (barcode, this_prob, this_diff)
             bc_comments = f"{best_bc}:{best_prob}:{best_diff}"
-            fq_repr = f"{name} {bc_comments}\n{seq}\n+\n{qual}\n"
+            fq_repr = f"{name} {bc_comments}\n{seq}\n+\n{qual}\n".encode("utf-8")
             if best_diff <= args.max_diff:
                 outfiles[best_bc].write(fq_repr)
             else: