mutations_caller_pipeline_aws 0.0.12 → 0.0.13
Sign up to get free protection for your applications and to get access to all the features.
data/bin/gatk_pipe_only_aws
CHANGED
@@ -44,7 +44,7 @@ options = { :bam_file_sorted => nil,
|
|
44
44
|
:vcf => nil,
|
45
45
|
:account => "",
|
46
46
|
:project => "",
|
47
|
-
:debug => 1
|
47
|
+
:debug => 1
|
48
48
|
}
|
49
49
|
|
50
50
|
optparse = OptionParser.new do |opts|
|
@@ -111,6 +111,7 @@ recal_bam = "#{random}_recal.bam"
|
|
111
111
|
options[:account] = options[:project] if options[:account].empty?
|
112
112
|
|
113
113
|
|
114
|
+
|
114
115
|
# Indexing
|
115
116
|
SamtoolsIndexing.call(bam_file,
|
116
117
|
job_prefix,
|
@@ -48,6 +48,8 @@ options = { :mutant_r1 => nil,
|
|
48
48
|
:account => "",
|
49
49
|
:project => "",
|
50
50
|
:debug => 1,
|
51
|
+
:cluster => false,
|
52
|
+
:coverage => false
|
51
53
|
}
|
52
54
|
|
53
55
|
optparse = OptionParser.new do |opts|
|
@@ -89,6 +91,14 @@ optparse = OptionParser.new do |opts|
|
|
89
91
|
options[:debug] = 5 if i
|
90
92
|
end
|
91
93
|
|
94
|
+
opts.on("-p","--cluster ", "On compute cluster (none AWS)" ) do |i|
|
95
|
+
options[:cluster] = true
|
96
|
+
end
|
97
|
+
|
98
|
+
opts.on("-o","--coverage ", "GATK Coverage summary" ) do |i|
|
99
|
+
options[:coverage] = true
|
100
|
+
end
|
101
|
+
|
92
102
|
opts.on_tail("-h", "--help", "Show this message") do
|
93
103
|
puts opts
|
94
104
|
exit
|
@@ -125,6 +135,7 @@ sai_file_fwd = "#{random}_fwd.sai"
|
|
125
135
|
sai_file_rev = "#{random}_rev.sai"
|
126
136
|
|
127
137
|
options[:account] = options[:project] if options[:account].empty?
|
138
|
+
options[:gatk] = "java -Xmx4g -jar #{options[:gatk]}" if options[:cluster]
|
128
139
|
|
129
140
|
# BWA : ALN
|
130
141
|
BwaCaller.call_aln(options[:mutant_r1],
|
@@ -219,3 +230,17 @@ GatkCaller.call(log_file,
|
|
219
230
|
job_prefix,
|
220
231
|
options[:account],
|
221
232
|
options[:debug])
|
233
|
+
|
234
|
+
|
235
|
+
# GATK : coverage
|
236
|
+
if options[:coverage]
|
237
|
+
outfile_prefix = options[:vcf].gsub(/.vcf/,"_coverage")
|
238
|
+
GatkCaller.coverage(
|
239
|
+
log_file,
|
240
|
+
options[:gatk],
|
241
|
+
options[:index_fa],
|
242
|
+
recal_bam,
|
243
|
+
job_prefix,
|
244
|
+
options[:account],
|
245
|
+
options[:debug])
|
246
|
+
end
|
@@ -3,7 +3,7 @@ class GatkCaller
|
|
3
3
|
# Genotyper
|
4
4
|
def self.call(log_dir, gatk, index_fa, read_bam, read_vcf, job_prefix, account, debug)
|
5
5
|
cmd = "echo 'starting GATK for mutant at ' `date` >> #{log_dir}
|
6
|
-
qsub -o #{log_dir} -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=
|
6
|
+
qsub -o #{log_dir} -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=4.5G -hold_jid recalibration_#{job_prefix} #{account}\
|
7
7
|
#{gatk} -l INFO -R #{index_fa} -T UnifiedGenotyper \
|
8
8
|
-I #{read_bam} \
|
9
9
|
-o #{read_vcf} \
|
@@ -12,10 +12,22 @@ class GatkCaller
|
|
12
12
|
system(cmd) if debug == 1
|
13
13
|
end
|
14
14
|
|
15
|
+
# INDEX is normal genom.fa
|
16
|
+
# Coverage Summary
|
17
|
+
def self.coverage(log_dir, gatk, index_fa, read_bam, outfile_prefix, job_prefix, account, debug)
|
18
|
+
cmd = "echo 'starting coverage GATK for mutant at ' `date` >> #{log_dir}
|
19
|
+
qsub -o #{log_dir} -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=4.5G -hold_jid recalibration_#{job_prefix} #{account}\
|
20
|
+
#{gatk} -R #{index_fa} -T DepthOfCoverage \
|
21
|
+
-I #{read_bam} \
|
22
|
+
-o #{outfile_prefix} "
|
23
|
+
puts cmd
|
24
|
+
system(cmd) if debug == 1
|
25
|
+
end
|
26
|
+
|
15
27
|
# Making recalibration table
|
16
28
|
def self.recalibrate_bam(log_dir ,gatk, index_fa, index_vcf, read_bam, recal_file, job_prefix, account, debug )
|
17
29
|
cmd = "echo 'starting recalibration table ' `date` >> #{log_dir}
|
18
|
-
qsub -o #{log_dir} -V -cwd -b y -N recalibration_table_#{job_prefix} -l h_vmem=
|
30
|
+
qsub -o #{log_dir} -V -cwd -b y -N recalibration_table_#{job_prefix} -l h_vmem=4.5G -hold_jid realignment_#{job_prefix} #{account} \
|
19
31
|
#{gatk} -knownSites #{index_vcf} -I #{read_bam} \
|
20
32
|
-R #{index_fa} -T CountCovariates \
|
21
33
|
-cov ReadGroupCovariate -cov QualityScoreCovariate -cov DinucCovariate \
|
@@ -28,7 +40,7 @@ class GatkCaller
|
|
28
40
|
# Using recalibration table
|
29
41
|
def self.table_calibration(log_dir, gatk, index_fa, read_bam, recal_bam, recal_file, job_prefix, account, debug)
|
30
42
|
cmd = "echo 'recalibrating bam_file at ' `date` >> #{log_dir}
|
31
|
-
qsub -V -o #{log_dir} -cwd -b y -N recalibration_#{job_prefix} -l h_vmem=
|
43
|
+
qsub -V -o #{log_dir} -cwd -b y -N recalibration_#{job_prefix} -l h_vmem=4.5G -hold_jid recalibration_table_#{job_prefix} #{account} \
|
32
44
|
#{gatk} \
|
33
45
|
-R #{index_fa} \
|
34
46
|
-I #{read_bam} \
|
@@ -42,7 +54,7 @@ class GatkCaller
|
|
42
54
|
# Preparation realignement
|
43
55
|
def self.prepare_realigne(log_dir, gatk, read_bam, index_fa, target_intervals, job_prefix, account, debug)
|
44
56
|
cmd = "echo 'preparing realignement at ' `date` >> #{log_dir}
|
45
|
-
qsub -o #{log_dir} -V -cwd -b y -N prep_realignment_#{job_prefix} -l h_vmem=
|
57
|
+
qsub -o #{log_dir} -V -cwd -b y -N prep_realignment_#{job_prefix} -l h_vmem=4.5G -hold_jid indexing_#{job_prefix} #{account}\
|
46
58
|
#{gatk} \
|
47
59
|
-I #{read_bam} \
|
48
60
|
-R #{index_fa} \
|
@@ -55,7 +67,7 @@ class GatkCaller
|
|
55
67
|
# Realignment
|
56
68
|
def self.realigne(log_dir, gatk, read_bam, index_fa, target_intervals, realigned_bam, job_prefix, account, debug)
|
57
69
|
cmd = "echo 'preparing realignement at ' `date` >> #{log_dir}
|
58
|
-
qsub -o #{log_dir} -V -cwd -b y -N realignment_#{job_prefix} -l h_vmem=
|
70
|
+
qsub -o #{log_dir} -V -cwd -b y -N realignment_#{job_prefix} -l h_vmem=4.5G -hold_jid prep_realignment_#{job_prefix} #{account} \
|
59
71
|
#{gatk} \
|
60
72
|
-I #{read_bam} \
|
61
73
|
-R #{index_fa} \
|
metadata
CHANGED
@@ -1,25 +1,35 @@
|
|
1
|
-
--- !ruby/object:Gem::Specification
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
2
|
name: mutations_caller_pipeline_aws
|
3
|
-
version: !ruby/object:Gem::Version
|
4
|
-
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
hash: 5
|
5
5
|
prerelease:
|
6
|
+
segments:
|
7
|
+
- 0
|
8
|
+
- 0
|
9
|
+
- 13
|
10
|
+
version: 0.0.13
|
6
11
|
platform: ruby
|
7
|
-
authors:
|
12
|
+
authors:
|
8
13
|
- Kaharina Hayer
|
9
14
|
autorequire:
|
10
15
|
bindir: bin
|
11
16
|
cert_chain: []
|
12
|
-
|
17
|
+
|
18
|
+
date: 2012-01-20 00:00:00 -05:00
|
19
|
+
default_executable:
|
13
20
|
dependencies: []
|
21
|
+
|
14
22
|
description: Using BWA to align and GATK to call the bases
|
15
|
-
email:
|
23
|
+
email:
|
16
24
|
- katharinaehayer@gmail.com
|
17
|
-
executables:
|
25
|
+
executables:
|
18
26
|
- mutations_caller_pipeline_aws
|
19
27
|
- gatk_pipe_only_aws
|
20
28
|
extensions: []
|
29
|
+
|
21
30
|
extra_rdoc_files: []
|
22
|
-
|
31
|
+
|
32
|
+
files:
|
23
33
|
- bin/gatk_pipe_only_aws
|
24
34
|
- bin/mutations_caller_pipeline_aws
|
25
35
|
- lib/mutations_caller_pipeline_aws.rb
|
@@ -28,28 +38,39 @@ files:
|
|
28
38
|
- lib/mutations_caller_pipeline_aws/gatk_caller.rb
|
29
39
|
- lib/mutations_caller_pipeline_aws/location_file.rb
|
30
40
|
- lib/mutations_caller_pipeline_aws/samtools_indexing.rb
|
41
|
+
has_rdoc: true
|
31
42
|
homepage: https://github.com/khayer/mutations_caller_pipeline_aws
|
32
43
|
licenses: []
|
44
|
+
|
33
45
|
post_install_message:
|
34
46
|
rdoc_options: []
|
35
|
-
|
47
|
+
|
48
|
+
require_paths:
|
36
49
|
- lib
|
37
|
-
required_ruby_version: !ruby/object:Gem::Requirement
|
50
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
38
51
|
none: false
|
39
|
-
requirements:
|
40
|
-
- -
|
41
|
-
- !ruby/object:Gem::Version
|
42
|
-
|
43
|
-
|
52
|
+
requirements:
|
53
|
+
- - ">="
|
54
|
+
- !ruby/object:Gem::Version
|
55
|
+
hash: 3
|
56
|
+
segments:
|
57
|
+
- 0
|
58
|
+
version: "0"
|
59
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
44
60
|
none: false
|
45
|
-
requirements:
|
46
|
-
- -
|
47
|
-
- !ruby/object:Gem::Version
|
48
|
-
|
61
|
+
requirements:
|
62
|
+
- - ">="
|
63
|
+
- !ruby/object:Gem::Version
|
64
|
+
hash: 3
|
65
|
+
segments:
|
66
|
+
- 0
|
67
|
+
version: "0"
|
49
68
|
requirements: []
|
69
|
+
|
50
70
|
rubyforge_project: mutations_caller_pipeline_aws
|
51
|
-
rubygems_version: 1.
|
71
|
+
rubygems_version: 1.6.2
|
52
72
|
signing_key:
|
53
73
|
specification_version: 3
|
54
74
|
summary: Call Mutations for files.fq
|
55
75
|
test_files: []
|
76
|
+
|