annotate phe/variant_filters/GTFilter.py @ 10:c2f8e7580133 draft

Uploaded
author ulfschaefer
date Mon, 21 Dec 2015 10:50:17 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
10
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
1 '''Filter VCF on GT filter parameter.
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
2
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
3 Created on 24 Sep 2015
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
4
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
5 @author: alex
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
6 '''
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
7
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
8 import argparse
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
9 import logging
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
10
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
11 from phe.variant_filters import PHEFilterBase
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
12
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
13
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
14 class UncallableGTFilter(PHEFilterBase):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
15 '''Filter uncallable genotypes'''
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
16
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
17 name = "UncallGT"
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
18 _default_threshold = None
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
19 parameter = "uncall_gt"
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
20
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
21 @classmethod
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
22 def customize_parser(self, parser):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
23 arg_name = self.parameter.replace("_", "-")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
24 parser.add_argument("--%s" % arg_name, type=str, default=self._default_threshold,
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
25 help="Filter sites below given GQ score (default: %s)" % self._default_threshold)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
26
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
27 def __init__(self, args):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
28 """Min Depth constructor."""
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
29 # This needs to happen first, because threshold is initialised here.
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
30 super(UncallableGTFilter, self).__init__(args)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
31
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
32 # Change the threshold to custom gq value.
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
33 self.threshold = self._default_threshold
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
34 if isinstance(args, argparse.Namespace):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
35 self.threshold = args.gq_score
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
36 elif isinstance(args, dict):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
37 try:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
38 self.threshold = str(args.get(self.parameter))
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
39 except (TypeError, ValueError):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
40 logging.error("Could not retrieve threshold from %s", args.get(self.parameter))
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
41 logging.error("This parameter requires to be a string!")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
42 self.threshold = None
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
43
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
44 def __call__(self, record):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
45 """Filter a :py:class:`vcf.model._Record`."""
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
46
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
47 if len(record.samples) > 1:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
48 logging.warn("More than 1 sample detected. Only first is considered.")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
49
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
50 try:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
51 record_gt = record.samples[0].data.GT
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
52 except AttributeError:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
53 logging.warn("Could not retrieve GQ score POS %i", record.POS)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
54 record_gt = None
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
55
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
56 if record_gt is None:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
57 return "./."
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
58 else:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
59 return None
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
60
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
61 def short_desc(self):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
62 short_desc = self.__doc__ or ''
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
63
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
64 if short_desc:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
65 short_desc = "%s (GT != ./. )" % (short_desc)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
66
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
67 return short_desc
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
68
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
69 def is_gap(self):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
70 return True
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
71
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
72 def is_n(self):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
73 return False