annotate phe/variant_filters/GQFilter.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 GQ parameter.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
2 Created on 24 Sep 2015
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
3
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
4 @author: alex
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
5 '''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
6
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
7 import argparse
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
8 import logging
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
9
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
10 from phe.variant_filters import PHEFilterBase
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
11
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
12
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
13 class GQFilter(PHEFilterBase):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
14 '''Filter sites by GQ score.'''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
15
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
16 name = "MinGQ"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
17 _default_threshold = 0
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
18 parameter = "gq_score"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
19
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
20 @classmethod
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
21 def customize_parser(self, parser):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
22 arg_name = self.parameter.replace("_", "-")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
23 parser.add_argument("--%s" % arg_name, type=int, default=self._default_threshold,
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
24 help="Filter sites below given GQ score (default: %s)" % self._default_threshold)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
25
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
26 def __init__(self, args):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
27 """Min Depth constructor."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
28 # This needs to happen first, because threshold is initialised here.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
29 super(GQFilter, self).__init__(args)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
30
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
31 # Change the threshold to custom gq value.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
32 self.threshold = self._default_threshold
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
33 if isinstance(args, argparse.Namespace):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
34 self.threshold = args.gq_score
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
35 elif isinstance(args, dict):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
36 try:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
37 self.threshold = int(args.get(self.parameter))
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
38 except TypeError:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
39 logging.error("Could not retrieve threshold from %s", args.get(self.parameter))
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
40 self.threshold = None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
41
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
42 def __call__(self, record):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
43 """Filter a :py:class:`vcf.model._Record`."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
44
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
45 if not record.is_snp:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
46 return None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
47
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
48 if len(record.samples) > 1:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
49 logging.warn("More than 1 sample detected. Only first is considered.")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
50
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
51 try:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
52 record_gq = record.samples[0].data.GQ
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
53 except AttributeError:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
54 logging.error("Could not retrieve GQ score POS %i", record.POS)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
55 record_gq = None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
56
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
57 if record_gq is None or record_gq < self.threshold:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
58 # FIXME: when record_gq is None, i,e, error, what do you do?
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
59 return record_gq or False
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
60 else:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
61 return None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
62
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
63 def short_desc(self):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
64 short_desc = self.__doc__ or ''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
65
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
66 if short_desc:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
67 short_desc = "%s (GQ > %s)" % (short_desc, self.threshold)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
68
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
69 return short_desc