diff phe/variant_filters/GQFilter.py @ 0:834a312c0114 draft

Uploaded

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/phe/variant_filters/GQFilter.py	Thu Dec 10 09:22:39 2015 -0500
@@ -0,0 +1,69 @@
+'''Filter VCF on GQ parameter.
+Created on 24 Sep 2015
+
+@author: alex
+'''
+
+import argparse
+import logging
+
+from phe.variant_filters import PHEFilterBase
+
+
+class GQFilter(PHEFilterBase):
+    '''Filter sites by GQ score.'''
+
+    name = "MinGQ"
+    _default_threshold = 0
+    parameter = "gq_score"
+
+    @classmethod
+    def customize_parser(self, parser):
+        arg_name = self.parameter.replace("_", "-")
+        parser.add_argument("--%s" % arg_name, type=int, default=self._default_threshold,
+                help="Filter sites below given GQ score (default: %s)" % self._default_threshold)
+
+    def __init__(self, args):
+        """Min Depth constructor."""
+        # This needs to happen first, because threshold is initialised here.
+        super(GQFilter, self).__init__(args)
+
+        # Change the threshold to custom gq value.
+        self.threshold = self._default_threshold
+        if isinstance(args, argparse.Namespace):
+            self.threshold = args.gq_score
+        elif isinstance(args, dict):
+            try:
+                self.threshold = int(args.get(self.parameter))
+            except TypeError:
+                logging.error("Could not retrieve threshold from %s", args.get(self.parameter))
+                self.threshold = None
+
+    def __call__(self, record):
+        """Filter a :py:class:`vcf.model._Record`."""
+
+        if not record.is_snp:
+            return None
+
+        if len(record.samples) > 1:
+            logging.warn("More than 1 sample detected. Only first is considered.")
+
+        try:
+            record_gq = record.samples[0].data.GQ
+        except AttributeError:
+            logging.error("Could not retrieve GQ score POS %i", record.POS)
+            record_gq = None
+
+        if record_gq is None or record_gq < self.threshold:
+            # FIXME: when record_gq is None, i,e, error, what do you do?
+            return record_gq or False
+        else:
+            return None
+
+    def short_desc(self):
+        short_desc = self.__doc__ or ''
+
+        if short_desc:
+            short_desc = "%s (GQ > %s)" % (short_desc, self.threshold)
+
+        return short_desc