cwl/preprocess/gvcf/src/cleanvcf.py

   1 # Copyright (C) The Lightning Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 #!/usr/bin/env python
   6
   7 from __future__ import print_function
   8 import sys
   9
  10 def is_header(line):
  11     """Check if a line is header."""
  12
  13     return line.startswith('#')
  14
  15 # FIELD index
  16 # CHROM 0, POS 1, REF 3
  17
  18 def main():
  19     previous_CHROM = ""
  20     previous_end_POS = 0
  21
  22     for line in sys.stdin:
  23         if not is_header(line):
  24             fields = line.split('\t')
  25             CHROM = fields[0]
  26             POS = int(fields[1])
  27             REF = fields[3]
  28             if CHROM == previous_CHROM:
  29                 if POS > previous_end_POS:
  30                     print(line, end='')
  31                     previous_end_POS = max(previous_end_POS, POS + len(REF) - 1)
  32             else:
  33                 print(line, end='')
  34                 previous_end_POS = POS + len(REF) - 1
  35             previous_CHROM = CHROM
  36         else:
  37             print(line, end='')
  38
  39 if __name__ == '__main__':
  40     main()