extra/fix_ob3.py

   1 """
   2 Fix off by 3 error in our database
   3 """
   4 from optparse import OptionParser
   5 import os
   6 import re
   7 import sys
   8
   9 from django.core.management import setup_environ
  10 from django.conf import settings
  11 setup_environ(settings)
  12
  13 import htsworkflow.frontend.samples.models as samples
  14
  15 def main(cmdline=None):
  16     parser = make_parser()
  17     opts, args = parser.parse_args(cmdline)
  18
  19     dry_run = not opts.run
  20     fix_ob3(dry_run=dry_run)
  21
  22     return 0
  23
  24 def make_parser():
  25     parser = OptionParser("%prog: fix off by 3 error that creeped in")
  26     parser.add_option("--run", default=False, action="store_true",
  27                       help="change the database")
  28     return parser
  29
  30 def fix_ob3(dry_run=True):
  31     libraries = samples.Library.objects.order_by('id')
  32
  33     mismatch = 0
  34     wrong_amp = 0
  35     wrong_amp_ids = set()
  36     for lib in libraries:
  37         if lib.amplified_from_sample is not None:
  38             amp_sample = lib.amplified_from_sample
  39             alt_sample = samples.Library.objects.get(pk=int(amp_sample.id)-3)
  40             if is_alt_sample_right(lib, alt_sample):
  41                 wrong_amp += 1
  42                 wrong_amp_ids.add(int(lib.id))
  43                 print "--- wrong lib ---"
  44                 display_names(lib, amp_sample, alt_sample)
  45                 if not dry_run:
  46                     lib.amplified_from_sample = alt_sample
  47                     lib.save()
  48             #elif lib_name != amp_sample.library_name:
  49             #    mismatch += 1
  50             #    print "--- didn't match ---"
  51             #    display_names(lib_name, lib, amp_sample, other_sample)
  52     print "-----"
  53     print "{0} mismatches".format(mismatch)
  54     print "{0} obviously wrong libs".format(wrong_amp)
  55     if len(wrong_amp_ids) > 0:
  56         print "    {0} - {1}".format(min(wrong_amp_ids), max(wrong_amp_ids))
  57
  58 def clean_lib_name(lib):
  59     """Strip trailing amplified marker character"""
  60     return re.sub(" *a$", "", lib.library_name)
  61
  62 def is_alt_sample_right(lib, alt_sample):
  63     """Check to see if the alt sample is the right sample
  64     """
  65     lib_name = clean_lib_name(lib)
  66     return lib_name == alt_sample.library_name
  67
  68 def display_names(lib, amp_from_sample, alt_from_sample):
  69     lib_name = clean_lib_name(lib)
  70     print "NonA: "+lib.id+"|"+lib_name+"|"
  71     print "   A: "+lib.id+"|"+lib.library_name+"|"
  72     print "AmpF: "+ amp_from_sample.id+"|"+amp_from_sample.library_name+"|"
  73     print "FixF: "+ alt_from_sample.id+"|"+alt_from_sample.library_name+"|"
  74
  75
  76 if __name__ == "__main__":
  77     sys.exit(main())