annotate phe/variant_filters/GTFilter.py @ 0:834a312c0114 draft

Uploaded
author ulfschaefer
date Thu, 10 Dec 2015 09:22:39 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
1 '''Filter VCF on GT filter parameter.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
2
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
3 Created on 24 Sep 2015
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
4
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
5 @author: alex
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
6 '''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
7
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
8 import argparse
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
9 import logging
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
10
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
11 from phe.variant_filters import PHEFilterBase
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
12
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
13
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
14 class UncallableGTFilter(PHEFilterBase):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
15 '''Filter uncallable genotypes'''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
16
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
17 name = "UncallGT"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
18 _default_threshold = None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
19 parameter = "uncall_gt"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
20
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
21 @classmethod
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
22 def customize_parser(self, parser):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
23 arg_name = self.parameter.replace("_", "-")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
24 parser.add_argument("--%s" % arg_name, type=str, default=self._default_threshold,
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
25 help="Filter sites below given GQ score (default: %s)" % self._default_threshold)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
26
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
27 def __init__(self, args):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
28 """Min Depth constructor."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
29 # This needs to happen first, because threshold is initialised here.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
30 super(UncallableGTFilter, self).__init__(args)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
31
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
32 # Change the threshold to custom gq value.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
33 self.threshold = self._default_threshold
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
34 if isinstance(args, argparse.Namespace):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
35 self.threshold = args.gq_score
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
36 elif isinstance(args, dict):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
37 try:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
38 self.threshold = str(args.get(self.parameter))
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
39 except TypeError:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
40 logging.error("Could not retrieve threshold from %s", args.get(self.parameter))
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
41 self.threshold = None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
42
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
43 def __call__(self, record):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
44 """Filter a :py:class:`vcf.model._Record`."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
45
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
46 if len(record.samples) > 1:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
47 logging.warn("More than 1 sample detected. Only first is considered.")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
48
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
49 try:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
50 record_gt = record.samples[0].data.GT
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
51 except AttributeError:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
52 logging.error("Could not retrieve GQ score POS %i", record.POS)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
53 record_gt = None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
54
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
55 if record_gt is None:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
56 return "./."
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
57 else:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
58 return None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
59
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
60 def short_desc(self):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
61 short_desc = self.__doc__ or ''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
62
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
63 if short_desc:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
64 short_desc = "%s (GT != ./. )" % (short_desc)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
65
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
66 return short_desc
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
67
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
68 def is_gap(self):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
69 return True
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
70
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
71 def is_n(self):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
72 return False