annotate phe/variant/GATKVariantCaller.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 '''
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
2 Created on 22 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 from collections import OrderedDict
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
7 import logging
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
8 import os
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
9 import subprocess
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
10
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
11 from phe.variant import VariantCaller
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
12
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
13
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
14 class GATKVariantCaller(VariantCaller):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
15 """Implemetation of the Broad institute's variant caller."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
16
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
17 name = "gatk"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
18 """Plain text name of the variant caller."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
19
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
20 _default_options = "--sample_ploidy 2 --genotype_likelihoods_model BOTH -rf BadCigar -out_mode EMIT_ALL_SITES -nt 1"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
21 """Default options for the variant caller."""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
22
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
23 def __init__(self, cmd_options=None):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
24 """Constructor"""
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
25 if cmd_options is None:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
26 cmd_options = self._default_options
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
27
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
28 super(GATKVariantCaller, self).__init__(cmd_options=cmd_options)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
29
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
30 self.last_command = None
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
31
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
32 def get_info(self, plain=False):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
33 d = {"name": "gatk", "version": self.get_version(), "command": self.last_command}
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
34
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
35 if plain:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
36 result = "GATK(%(version)s): %(command)s" % d
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
37 else:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
38 result = OrderedDict(d)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
39
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
40 return result
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
41
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
42 def get_version(self):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
43
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
44 p = subprocess.Popen(["java", "-jar", os.environ["GATK_JAR"], "-version"], stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
45 (output, _) = p.communicate()
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
46
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
47 # last character is EOL.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
48 version = output.split("\n")[-2]
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
49
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
50 return version
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
51
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
52 def make_vcf(self, *args, **kwargs):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
53 ref = kwargs.get("ref")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
54 bam = kwargs.get("bam")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
55
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
56 if kwargs.get("vcf_file") is None:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
57 kwargs["vcf_file"] = "variants.vcf"
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
58
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
59 opts = {"ref": os.path.abspath(ref),
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
60 "bam": os.path.abspath(bam),
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
61 "gatk_jar": os.environ["GATK_JAR"],
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
62 "all_variants_file": os.path.abspath(kwargs.get("vcf_file")),
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
63 "extra_cmd_options": self.cmd_options}
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
64
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
65 # if not self.create_aux_files(ref):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
66 # logging.warn("Auxiliary files were not created.")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
67 # return False
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
68
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
69 # Call variants
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
70 # FIXME: Sample ploidy = 2?
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
71 os.environ["GATK_JAR"]
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
72 cmd = "java -XX:+UseSerialGC -jar %(gatk_jar)s -T UnifiedGenotyper -R %(ref)s -I %(bam)s -o %(all_variants_file)s %(extra_cmd_options)s" % opts
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
73 success = os.system(cmd)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
74
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
75 if success != 0:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
76 logging.warn("Calling variants returned non-zero exit status.")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
77 return False
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
78
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
79 self.last_command = cmd
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
80
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
81 return True
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
82
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
83 def create_aux_files(self, ref):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
84 """Create auxiliary files needed for this variant.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
85
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
86 Tools needed: samtools and picard tools. Picard tools is a Java
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
87 library that can be defined using environment variable: PICARD_JAR
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
88 specifying path to picard.jar or PICARD_TOOLS_PATH specifying path
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
89 to the directory where separate jars are (older version before jars
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
90 were merged into a single picard.jar).
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
91 Parameters:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
92 -----------
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
93 ref: str
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
94 Path to the reference file.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
95
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
96 Returns:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
97 --------
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
98 bool:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
99 True if auxiliary files were created, False otherwise.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
100 """
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
101
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
102 ref_name, _ = os.path.splitext(ref)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
103
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
104 success = os.system("samtools faidx %s" % ref)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
105
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
106 if success != 0:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
107 logging.warn("Fasta index could not be created.")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
108 return False
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
109
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
110 d = {"ref": ref, "ref_name": ref_name}
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
111
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
112 if os.environ.get("PICARD_TOOLS_PATH"):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
113 d["picard_tools_path"] = os.path.join(os.environ["PICARD_TOOLS_PATH"], "CreateSequenceDictionary.jar")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
114 elif os.environ.get("PICARD_JAR"):
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
115 # This is used in newer version of PICARD tool where multiple
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
116 # jars were merged into a single jar file.
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
117 d["picard_tools_path"] = "%s %s" % (os.environ["PICARD_JAR"], "CreateSequenceDictionary")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
118 else:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
119 logging.error("Picard tools are not present in the path.")
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
120 return False
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
121
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
122 success = os.system("java -jar %(picard_tools_path)s R=%(ref)s O=%(ref_name)s.dict" % d)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
123
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
124 if success != 0:
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
125 logging.warn("Dictionary for the %s reference could not be created", ref)
834a312c0114 Uploaded
ulfschaefer
parents:
diff changeset
126 return False