annotate phe/variant_filters/DepthFilter.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 depth of coverage.
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 import argparse
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
8 import logging
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
9
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
10 from phe.variant_filters import PHEFilterBase
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
11
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
12
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
13 class DepthFilter(PHEFilterBase):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
14 """Filter sites by depth."""
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
15
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
16 name = "MinDepth"
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
17 _default_threshold = 5
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
18 parameter = "min_depth"
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
19
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
20 @classmethod
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
21 def customize_parser(self, parser):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
22 arg_name = self.parameter.replace("_", "-")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
23 parser.add_argument("--" % arg_name, type=int, default=self._default_threshold,
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
24 help="Filter sites below minimum depth (default: %s)" % self._default_threshold)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
25
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
26 def __init__(self, args):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
27 """Min Depth constructor."""
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
28 # This needs to happen first, because threshold is initialised here.
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
29 super(DepthFilter, self).__init__(args)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
30
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
31 # Change the threshold to custom dp value.
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
32 self.threshold = self._default_threshold
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
33 if isinstance(args, argparse.Namespace):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
34 self.threshold = args.min_depth
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
35 elif isinstance(args, dict):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
36 try:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
37 self.threshold = int(args.get(self.parameter))
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
38 except (TypeError, ValueError):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
39 logging.error("Could not retrieve integer threshold from %s", args.get(self.parameter))
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
40 raise Exception("Could not create depth filter from parameters: %s" % args)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
41
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
42 def __call__(self, record):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
43 """Filter a :py:class:`vcf.model._Record`."""
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
44
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
45 if len(record.samples) > 1:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
46 logging.warn("Currently we only filter VCFs with 1 sample. Only first sample will be used.")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
47 logging.warn("This parameter requires to be an integer!")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
48
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
49 try:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
50 record_dp = record.samples[0].data.DP
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
51 except AttributeError:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
52 record_dp = None
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
53
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
54 if record_dp is None:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
55 # logging.debug("Falling back to INFO DP")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
56 record_dp = record.INFO.get("DP")
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
57
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
58 if record_dp is None or record_dp < self.threshold:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
59 return record_dp or False
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
60 else:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
61 return None
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
62
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
63 def short_desc(self):
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
64 short_desc = self.__doc__ or ''
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
65
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
66 if short_desc:
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
67 short_desc = "%s (DP > %i)" % (short_desc, self.threshold)
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
68
c2f8e7580133 Uploaded
ulfschaefer
parents:
diff changeset
69 return short_desc