mutations_caller_pipeline_aws 0.0.15 → 0.0.16
Sign up to get free protection for your applications and to get access to all the features.
- data/bin/gatk_pipe_only_aws +17 -3
- data/bin/mutations_caller_pipeline_aws +133 -29
- data/lib/mutations_caller_pipeline_aws.rb +1 -0
- data/lib/mutations_caller_pipeline_aws/bwa_caller.rb +7 -8
- data/lib/mutations_caller_pipeline_aws/gatk_caller.rb +28 -33
- data/lib/mutations_caller_pipeline_aws/picard_caller.rb +28 -0
- metadata +21 -41
data/bin/gatk_pipe_only_aws
CHANGED
@@ -4,6 +4,13 @@ require 'optparse'
|
|
4
4
|
require 'rubygems'
|
5
5
|
require 'yaml'
|
6
6
|
|
7
|
+
###
|
8
|
+
# NOT FUNTIONAL ANYMORE
|
9
|
+
###
|
10
|
+
|
11
|
+
|
12
|
+
|
13
|
+
|
7
14
|
usage =<<EOF
|
8
15
|
_________________________________________________________________________________________________
|
9
16
|
|
@@ -11,6 +18,7 @@ ________________________________________________________________________________
|
|
11
18
|
-b sorted_bam_file
|
12
19
|
-c config.yml -v raw_vcf_file
|
13
20
|
[-a account || -p project]
|
21
|
+
-s SampleSheet.csv
|
14
22
|
|
15
23
|
_________________________________________________________________________________________________
|
16
24
|
|
@@ -44,7 +52,8 @@ options = { :bam_file_sorted => nil,
|
|
44
52
|
:vcf => nil,
|
45
53
|
:account => "",
|
46
54
|
:project => "",
|
47
|
-
:debug => 1
|
55
|
+
:debug => 1,
|
56
|
+
:samplesheet => nil
|
48
57
|
}
|
49
58
|
|
50
59
|
optparse = OptionParser.new do |opts|
|
@@ -74,6 +83,10 @@ optparse = OptionParser.new do |opts|
|
|
74
83
|
options[:debug] = 5 if i
|
75
84
|
end
|
76
85
|
|
86
|
+
opts.on("-s", "--sampleSheet DIR", :REQUIRED, String, "SampleSheet.csv") do |i|
|
87
|
+
options[:samplesheet] = i
|
88
|
+
end
|
89
|
+
|
77
90
|
opts.on_tail("-h", "--help", "Show this message") do
|
78
91
|
puts opts
|
79
92
|
exit
|
@@ -82,7 +95,8 @@ end
|
|
82
95
|
|
83
96
|
begin
|
84
97
|
optparse.parse!
|
85
|
-
mandatory = [:bam_file_sorted, :index_prefix, :annotation_file,
|
98
|
+
mandatory = [:samplesheet, :bam_file_sorted, :index_prefix, :annotation_file,
|
99
|
+
:bwa, :samtools, :gatk, :vcf, :index_vcf, :index_fa]
|
86
100
|
missing = mandatory.select{ |param| options[param].nil? }
|
87
101
|
if !missing.empty?
|
88
102
|
puts "\nMissing options given or missing in config_file: \n\t#{missing.join(",\n\t")}"
|
@@ -112,7 +126,7 @@ options[:account] = options[:project] if options[:account].empty?
|
|
112
126
|
|
113
127
|
|
114
128
|
|
115
|
-
# Indexing
|
129
|
+
# Indexing + Mark Duplicates
|
116
130
|
SamtoolsIndexing.call(bam_file,
|
117
131
|
job_prefix,
|
118
132
|
options[:account],
|
@@ -3,6 +3,14 @@ require 'mutations_caller_pipeline_aws'
|
|
3
3
|
require 'optparse'
|
4
4
|
require 'rubygems'
|
5
5
|
require 'yaml'
|
6
|
+
require 'csv'
|
7
|
+
=begin
|
8
|
+
* Name: Mutations Caller Pipeline (AWS)
|
9
|
+
* Pipeline combining bwa with GATK2
|
10
|
+
* Author: Katharina Hayer
|
11
|
+
* Date: 8/8/2012
|
12
|
+
* License: GNU General Public License (GPL-2.0)
|
13
|
+
=end
|
6
14
|
|
7
15
|
usage =<<EOF
|
8
16
|
_________________________________________________________________________________________________
|
@@ -12,6 +20,8 @@ ________________________________________________________________________________
|
|
12
20
|
[-w wildtype_r1.fq -x wildtype_r2.fq]
|
13
21
|
-c config.yml -v raw_vcf_file
|
14
22
|
[-a account || -p project]
|
23
|
+
-s SampleSheet.csv
|
24
|
+
-f dbsnp_file
|
15
25
|
_________________________________________________________________________________________________
|
16
26
|
|
17
27
|
#{$0} ...
|
@@ -20,12 +30,12 @@ ________________________________________________________________________________
|
|
20
30
|
Also you should have the indices for bwa and GATK prepared.
|
21
31
|
NOTE: Only paired end reads are supported!
|
22
32
|
|
23
|
-
+++ C L U S T E R V E R S I O N - 0.0.
|
33
|
+
+++ C L U S T E R V E R S I O N - 0.0.15 +++
|
24
34
|
|
25
35
|
config.yml should look like this:
|
26
36
|
# config.yml
|
27
|
-
|
28
|
-
|
37
|
+
bwa_prefix: "path/to/prefix"
|
38
|
+
picard_tools: "path/to/picard_tools/"
|
29
39
|
bwa: "path/to/bwa"
|
30
40
|
samtools: "path/to/samtools"
|
31
41
|
gatk: "path/to/GenomeAnalysisTK.jar"
|
@@ -49,7 +59,8 @@ options = { :mutant_r1 => nil,
|
|
49
59
|
:project => "",
|
50
60
|
:debug => 1,
|
51
61
|
:cluster => false,
|
52
|
-
:coverage => false
|
62
|
+
:coverage => false,
|
63
|
+
:samplesheet => nil
|
53
64
|
}
|
54
65
|
|
55
66
|
optparse = OptionParser.new do |opts|
|
@@ -91,7 +102,7 @@ optparse = OptionParser.new do |opts|
|
|
91
102
|
options[:debug] = 5 if i
|
92
103
|
end
|
93
104
|
|
94
|
-
opts.on("-
|
105
|
+
opts.on("-k","--cluster ", "On compute cluster (none AWS)" ) do |i|
|
95
106
|
options[:cluster] = true if i
|
96
107
|
end
|
97
108
|
|
@@ -99,6 +110,19 @@ optparse = OptionParser.new do |opts|
|
|
99
110
|
options[:coverage] = true if i
|
100
111
|
end
|
101
112
|
|
113
|
+
opts.on("-s", "--sampleSheet DIR", :REQUIRED, String, "SampleSheet.csv") do |i|
|
114
|
+
options[:samplesheet] = i
|
115
|
+
end
|
116
|
+
|
117
|
+
opts.on("-f", "--dbsnp_file DIR", :REQUIRED, String, "dbsnp.vcf") do |i|
|
118
|
+
options[:dbsnp_file] = i
|
119
|
+
end
|
120
|
+
|
121
|
+
opts.on("-b", "--sample_name Name", :REQUIRED, String, "Sample name / SampleID") do |i|
|
122
|
+
options[:sample_name] = i
|
123
|
+
end
|
124
|
+
|
125
|
+
|
102
126
|
opts.on_tail("-h", "--help", "Show this message") do
|
103
127
|
puts opts
|
104
128
|
exit
|
@@ -107,7 +131,8 @@ end
|
|
107
131
|
|
108
132
|
begin
|
109
133
|
optparse.parse!
|
110
|
-
mandatory = [:
|
134
|
+
mandatory = [:dbsnp_file,:picard_tools, :samplesheet, :mutant_r1, :mutant_r2, :bwa_prefix,
|
135
|
+
:bwa, :samtools, :gatk, :vcf, :index_vcf, :sample_name, :index_fa]
|
111
136
|
missing = mandatory.select{ |param| options[param].nil? }
|
112
137
|
if !missing.empty?
|
113
138
|
puts "\nMissing options given or missing in config_file: \n\t#{missing.join(",\n\t")}"
|
@@ -120,41 +145,117 @@ rescue OptionParser::InvalidOption, OptionParser::MissingArgument
|
|
120
145
|
exit
|
121
146
|
end
|
122
147
|
|
148
|
+
# Methods
|
149
|
+
def execute(cmd)
|
150
|
+
puts cmd
|
151
|
+
status = system(cmd)
|
152
|
+
raise cmd unless status
|
153
|
+
end
|
154
|
+
|
155
|
+
def run(options)
|
156
|
+
# tmp files for output
|
157
|
+
job_number = options[:job_number]
|
158
|
+
bam_file = "aligned_#{job_number}.bam"
|
159
|
+
bam_file_sorted = "sorted_#{job_number}.bam"
|
160
|
+
bam_file_sorted_dublicates = options[:step_one_bam_file]
|
161
|
+
dublicate_metrcis = "GATK_files/dublicate.metrics"
|
162
|
+
job_prefix = "#{job_number}"
|
163
|
+
log_file = "log/#{options[:sample_name]}.log"
|
164
|
+
|
165
|
+
cmd = PicardCaller.convert(options[:sam_file], bam_file, options[:picard_tools],
|
166
|
+
log_file, job_prefix, options[:account])
|
167
|
+
|
168
|
+
if options[:debug] == 1
|
169
|
+
execute(cmd)
|
170
|
+
else
|
171
|
+
puts cmd
|
172
|
+
end
|
173
|
+
|
174
|
+
cmd = PicardCaller.rg_and_sorting(bam_file, bam_file_sorted, options[:picard_tools],
|
175
|
+
options[:library], options[:index], options[:sample_name],
|
176
|
+
log_file, options[:id], job_prefix, options[:account])
|
177
|
+
|
178
|
+
if options[:debug] == 1
|
179
|
+
execute(cmd)
|
180
|
+
else
|
181
|
+
puts cmd
|
182
|
+
end
|
183
|
+
|
184
|
+
cmd = PicardCaller.mark_dublicates(bam_file_sorted, bam_file_sorted_dublicates,
|
185
|
+
dublicate_metrcis, options[:picard_tools], log_file, job_prefix,
|
186
|
+
options[:account])
|
187
|
+
|
188
|
+
if options[:debug] == 1
|
189
|
+
execute(cmd)
|
190
|
+
else
|
191
|
+
puts cmd
|
192
|
+
end
|
193
|
+
|
194
|
+
cmd = PicardCaller.build_index(bam_file_sorted_dublicates,
|
195
|
+
options[:picard_tools], log_file, job_prefix, options[:account])
|
196
|
+
|
197
|
+
if options[:debug] == 1
|
198
|
+
execute(cmd)
|
199
|
+
else
|
200
|
+
puts cmd
|
201
|
+
end
|
202
|
+
end
|
203
|
+
|
204
|
+
|
123
205
|
# pipeline starts here
|
124
206
|
|
207
|
+
# get information from sample sheet
|
208
|
+
CSV.foreach(options[:samplesheet],{:headers => :first_row}) do |row|
|
209
|
+
if sample_name = row["SampleID"]
|
210
|
+
options[:index] = row["Index"]
|
211
|
+
lane = row["Lane"]
|
212
|
+
sample_project = row["SampleProject"]
|
213
|
+
options[:id] = "#{sample_project}_#{lane}_#{sample_name}"
|
214
|
+
options[:library] = row["FCID"]
|
215
|
+
end
|
216
|
+
end
|
217
|
+
|
218
|
+
# Create Dir to not overcluster output folder
|
219
|
+
Dir.mkdir("GATK_files") unless File.exists?("GATK_files")
|
220
|
+
Dir.mkdir("log") unless File.exists?("log")
|
221
|
+
|
125
222
|
# tmp files for output
|
126
223
|
random = (rand*1000000).floor.to_s
|
127
|
-
|
224
|
+
sample_name = options[:sample_name]
|
225
|
+
options[:sam_file] = "#{sample_name}_#{random}.sam"
|
226
|
+
options[:step_one_bam_file] = "#{sample_name}_#{random}.bam"
|
227
|
+
bam_file = "mutant_#{random}.bam"
|
128
228
|
job_prefix = "#{random}"
|
129
|
-
|
130
|
-
|
229
|
+
options[:job_number] = job_prefix
|
230
|
+
log_file = "log/#{sample_name}.log"
|
231
|
+
target_intervals = "GATK_files/target.intervals"
|
131
232
|
realigned_bam = "#{random}_realigned.bam"
|
132
|
-
recal_file = "
|
133
|
-
recal_bam = "#{
|
233
|
+
recal_file = "GATK_files/recal.grp"
|
234
|
+
recal_bam = "#{sample_name}.bam"
|
134
235
|
sai_file_fwd = "#{random}_fwd.sai"
|
135
236
|
sai_file_rev = "#{random}_rev.sai"
|
136
237
|
|
137
238
|
options[:account] = options[:project] if options[:account].empty?
|
138
|
-
options[:gatk] = "java -
|
239
|
+
#options[:gatk] = "java -Xmx8g -jar #{options[:gatk]}"
|
139
240
|
|
140
241
|
# BWA : ALN
|
141
242
|
BwaCaller.call_aln(options[:mutant_r1],
|
142
|
-
options[:
|
243
|
+
options[:bwa_prefix],
|
143
244
|
sai_file_fwd,
|
144
245
|
log_file,
|
145
246
|
options[:bwa],
|
146
247
|
job_prefix,
|
147
248
|
options[:account],
|
148
|
-
options[:debug])
|
249
|
+
options[:debug],"fwd")
|
149
250
|
|
150
251
|
BwaCaller.call_aln(options[:mutant_r2],
|
151
|
-
options[:
|
252
|
+
options[:bwa_prefix],
|
152
253
|
sai_file_rev,
|
153
254
|
log_file,
|
154
255
|
options[:bwa],
|
155
256
|
job_prefix,
|
156
257
|
options[:account],
|
157
|
-
options[:debug])
|
258
|
+
options[:debug],"rev")
|
158
259
|
|
159
260
|
|
160
261
|
# BWA : First step mapping reads to reference
|
@@ -162,8 +263,8 @@ BwaCaller.call_paired_end(options[:mutant_r1],
|
|
162
263
|
options[:mutant_r2],
|
163
264
|
sai_file_fwd,
|
164
265
|
sai_file_rev,
|
165
|
-
|
166
|
-
options[:
|
266
|
+
options[:sam_file],
|
267
|
+
options[:bwa_prefix],
|
167
268
|
log_file,
|
168
269
|
options[:bwa],
|
169
270
|
options[:samtools],
|
@@ -172,27 +273,30 @@ BwaCaller.call_paired_end(options[:mutant_r1],
|
|
172
273
|
options[:debug])
|
173
274
|
|
174
275
|
|
175
|
-
# Indexing
|
176
|
-
|
177
|
-
|
178
|
-
|
179
|
-
|
180
|
-
options[:
|
181
|
-
|
276
|
+
# Indexing + Marking Duplicates
|
277
|
+
run(options)
|
278
|
+
#bam_file = bam_file + ".bam"
|
279
|
+
#SamtoolsIndexing.call(bam_file,
|
280
|
+
# job_prefix,
|
281
|
+
# options[:account],
|
282
|
+
# options[:debug],
|
283
|
+
# log_file)
|
182
284
|
|
183
285
|
# Realigne
|
184
286
|
GatkCaller.prepare_realigne(log_file,
|
185
287
|
options[:gatk],
|
186
|
-
|
288
|
+
options[:step_one_bam_file],
|
187
289
|
options[:index_fa],
|
188
290
|
target_intervals,
|
189
291
|
job_prefix,
|
190
292
|
options[:account],
|
293
|
+
options[:dbsnp_file],
|
191
294
|
options[:debug])
|
192
295
|
|
296
|
+
|
193
297
|
GatkCaller.realigne(log_file,
|
194
298
|
options[:gatk],
|
195
|
-
|
299
|
+
options[:step_one_bam_file],
|
196
300
|
options[:index_fa],
|
197
301
|
target_intervals,
|
198
302
|
realigned_bam,
|
@@ -204,11 +308,11 @@ GatkCaller.realigne(log_file,
|
|
204
308
|
GatkCaller.recalibrate_bam( log_file,
|
205
309
|
options[:gatk],
|
206
310
|
options[:index_fa],
|
207
|
-
options[:index_vcf],
|
208
311
|
realigned_bam,
|
209
312
|
recal_file,
|
210
313
|
job_prefix,
|
211
314
|
options[:account],
|
315
|
+
options[:dbsnp_file],
|
212
316
|
options[:debug] )
|
213
317
|
|
214
318
|
GatkCaller.table_calibration(log_file,
|
@@ -225,11 +329,11 @@ GatkCaller.table_calibration(log_file,
|
|
225
329
|
GatkCaller.call(log_file,
|
226
330
|
options[:gatk],
|
227
331
|
options[:index_fa],
|
228
|
-
options[:index_vcf],
|
229
332
|
recal_bam,
|
230
333
|
options[:vcf],
|
231
334
|
job_prefix,
|
232
335
|
options[:account],
|
336
|
+
options[:dbsnp_file],
|
233
337
|
options[:debug])
|
234
338
|
|
235
339
|
|
@@ -1,6 +1,7 @@
|
|
1
1
|
require 'mutations_caller_pipeline_aws/bwa_caller'
|
2
2
|
require 'mutations_caller_pipeline_aws/gatk_caller'
|
3
3
|
require 'mutations_caller_pipeline_aws/samtools_indexing'
|
4
|
+
require 'mutations_caller_pipeline_aws/picard_caller'
|
4
5
|
|
5
6
|
class MutationsCallerPipelineAws
|
6
7
|
def self.hi
|
@@ -1,18 +1,17 @@
|
|
1
1
|
class BwaCaller
|
2
2
|
|
3
|
-
def self.call_paired_end(r1, r2, sai1, sai2,
|
3
|
+
def self.call_paired_end(r1, r2, sai1, sai2, sam_file, index, log_file, bwa, samtools, job_prefix,account, debug)
|
4
4
|
dummy = "\\\\\\"
|
5
|
-
cmd = "qsub -o #{log_file} -hold_jid bwa_aln_#{job_prefix} -V -cwd -b y -N bwa_#{job_prefix} -l h_vmem=6G
|
6
|
-
#{bwa} sampe
|
7
|
-
#{sai1} #{sai2} #{r1} #{r2}
|
8
|
-
\\| #{samtools} view -Su - \\| #{samtools} sort - #{out_file}"
|
5
|
+
cmd = "qsub -o #{log_file} -e #{log_file}_bwa_sampe_errors -hold_jid bwa_aln_#{job_prefix} -V -cwd -b y -N bwa_#{job_prefix} -l h_vmem=6G #{account}\
|
6
|
+
#{bwa} sampe #{index} \
|
7
|
+
#{sai1} #{sai2} #{r1} #{r2} -f #{sam_file}"
|
9
8
|
puts cmd
|
10
9
|
system('bash','-c', cmd) if debug == 1
|
11
10
|
end
|
12
11
|
|
13
|
-
def self.call_aln(read, index, out_file, log_file, bwa, job_prefix, account,debug)
|
14
|
-
cmd = "qsub -o #{log_file} -V -cwd -b y -N bwa_aln_#{job_prefix} -l h_vmem=4G #{account} \
|
15
|
-
#{bwa} aln -f #{out_file} #{index} #{read} "
|
12
|
+
def self.call_aln(read, index, out_file, log_file, bwa, job_prefix, account,debug,direction)
|
13
|
+
cmd = "qsub -pe DJ 8 -o #{log_file} -e #{log_file}_bwa_aln_errors_#{direction} -V -cwd -b y -N bwa_aln_#{job_prefix} -l h_vmem=4G #{account} \
|
14
|
+
#{bwa} aln -t 8 -f #{out_file} #{index} #{read} "
|
16
15
|
puts cmd
|
17
16
|
system(cmd) if debug == 1
|
18
17
|
end
|
@@ -1,13 +1,11 @@
|
|
1
1
|
class GatkCaller
|
2
2
|
# INDEX is normal genom.fa
|
3
3
|
# Genotyper
|
4
|
-
def self.call(log_dir, gatk, index_fa,
|
5
|
-
cmd = "
|
6
|
-
|
7
|
-
#{
|
8
|
-
-
|
9
|
-
-D #{dbSNP}
|
10
|
-
-o #{read_vcf} \
|
4
|
+
def self.call(log_dir, gatk, index_fa, read_bam, read_vcf, job_prefix, account,dbsnp_file, debug)
|
5
|
+
cmd = "qsub -pe DJ 4 -o #{log_dir} -e #{log_dir}_genotyper_errors -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=6G -hold_jid recalibration_#{job_prefix} #{account}\
|
6
|
+
java -Xmx6g -jar #{gatk} -l INFO -R #{index_fa} -T UnifiedGenotyper \
|
7
|
+
-I #{read_bam} --dbsnp #{dbsnp_file} \
|
8
|
+
-o #{read_vcf} -nt 4 --max_alternate_alleles 8 \
|
11
9
|
--genotype_likelihoods_model BOTH"
|
12
10
|
puts cmd
|
13
11
|
system(cmd) if debug == 1
|
@@ -15,49 +13,46 @@ class GatkCaller
|
|
15
13
|
|
16
14
|
# INDEX is normal genom.fa
|
17
15
|
# Coverage Summary
|
16
|
+
# parallel not possible yet (1.6-13-g91f02df)
|
18
17
|
def self.coverage(log_dir, gatk, index_fa, read_bam, outfile_prefix, job_prefix, account, debug)
|
19
|
-
cmd = "
|
20
|
-
|
21
|
-
|
22
|
-
-
|
23
|
-
-o #{outfile_prefix} "
|
18
|
+
cmd = "qsub -o #{log_dir} -e #{log_dir}_coverage_errors -V -cwd -b y -N coverage_#{job_prefix} -l h_vmem=7G -hold_jid recalibration_#{job_prefix} #{account}\
|
19
|
+
java -Xmx6g -jar #{gatk} -R #{index_fa} -T DepthOfCoverage \
|
20
|
+
-I #{read_bam} --omitDepthOutputAtEachBase \
|
21
|
+
-o #{outfile_prefix} --omitIntervalStatistics --omitLocusTable"
|
24
22
|
puts cmd
|
25
23
|
system(cmd) if debug == 1
|
26
24
|
end
|
27
25
|
|
28
26
|
# Making recalibration table
|
29
|
-
def self.recalibrate_bam(log_dir ,gatk, index_fa,
|
30
|
-
cmd = "
|
31
|
-
|
32
|
-
|
33
|
-
-
|
34
|
-
-cov ReadGroupCovariate -cov QualityScoreCovariate -cov DinucCovariate \
|
35
|
-
-cov CycleCovariate \
|
36
|
-
-recalFile #{recal_file}"
|
27
|
+
def self.recalibrate_bam(log_dir ,gatk, index_fa, read_bam, recal_file, job_prefix, account, dbsnp_file, debug )
|
28
|
+
cmd = "qsub -o #{log_dir} -e #{log_dir}_recalibrate_errors -V -cwd -b y -N recalibration_table_#{job_prefix} -l h_vmem=7G -hold_jid realignment_#{job_prefix} #{account} \
|
29
|
+
java -Xmx6g -jar #{gatk} -knownSites #{dbsnp_file} -I #{read_bam} \
|
30
|
+
-R #{index_fa} -T BaseRecalibrator \
|
31
|
+
-o #{recal_file}"
|
37
32
|
puts cmd
|
38
33
|
system(cmd) if debug == 1
|
39
34
|
end
|
40
35
|
|
41
36
|
# Using recalibration table
|
37
|
+
# parallel not possible yet (1.6-13-g91f02df)
|
42
38
|
def self.table_calibration(log_dir, gatk, index_fa, read_bam, recal_bam, recal_file, job_prefix, account, debug)
|
43
|
-
cmd = "
|
44
|
-
|
45
|
-
#{gatk} \
|
39
|
+
cmd = "qsub -V -o #{log_dir} -e #{log_dir}_prep_recal_errors -cwd -b y -N recalibration_#{job_prefix} -l h_vmem=7G -hold_jid recalibration_table_#{job_prefix} #{account} \
|
40
|
+
java -Xmx6g -jar #{gatk} \
|
46
41
|
-R #{index_fa} \
|
47
42
|
-I #{read_bam} \
|
48
|
-
-T
|
43
|
+
-T PrintReads \
|
49
44
|
-o #{recal_bam} \
|
50
|
-
-
|
45
|
+
-BQSR #{recal_file}"
|
51
46
|
puts cmd
|
52
47
|
system(cmd) if debug == 1
|
53
48
|
end
|
54
49
|
|
55
50
|
# Preparation realignement
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
#{gatk} \
|
60
|
-
-I #{read_bam} \
|
51
|
+
|
52
|
+
def self.prepare_realigne(log_dir, gatk, read_bam, index_fa, target_intervals, job_prefix, account, dbsnp_file, debug)
|
53
|
+
cmd = "qsub -pe DJ 4 -o #{log_dir} -e #{log_dir}_prep_realign_errors -V -cwd -b y -N prep_realignment_#{job_prefix} -l h_vmem=6G -hold_jid index_#{job_prefix} #{account}\
|
54
|
+
java -Xmx6g -jar #{gatk} -nt 4 \
|
55
|
+
-I #{read_bam} --known #{dbsnp_file} \
|
61
56
|
-R #{index_fa} \
|
62
57
|
-T RealignerTargetCreator \
|
63
58
|
-o #{target_intervals}"
|
@@ -66,10 +61,10 @@ class GatkCaller
|
|
66
61
|
end
|
67
62
|
|
68
63
|
# Realignment
|
64
|
+
# parallel not possible yet (1.6-13-g91f02df)
|
69
65
|
def self.realigne(log_dir, gatk, read_bam, index_fa, target_intervals, realigned_bam, job_prefix, account, debug)
|
70
|
-
cmd = "
|
71
|
-
|
72
|
-
#{gatk} \
|
66
|
+
cmd = "qsub -o #{log_dir} -e #{log_dir}_realign_errors -V -cwd -b y -N realignment_#{job_prefix} -l h_vmem=7G -hold_jid prep_realignment_#{job_prefix} #{account} \
|
67
|
+
java -Xmx6g -jar #{gatk} \
|
73
68
|
-I #{read_bam} \
|
74
69
|
-R #{index_fa} \
|
75
70
|
-T IndelRealigner \
|
@@ -0,0 +1,28 @@
|
|
1
|
+
class PicardCaller
|
2
|
+
#converter = "java -jar ~/Downloads/picard-tools-1.56/picard-tools-1.56/SamFormatConverter.jar I=WT_aligned_sorted_rg.bam O=tmp.sam VALIDATION_STRINGENCY=LENIENT"
|
3
|
+
def self.convert(sam_file, bam_file, picard_tools, log_file, job_prefix, account)
|
4
|
+
cmd = "qsub -o #{log_file} -e #{log_file}_conversion_errors -V -cwd -b y -hold_jid bwa_#{job_prefix} -N convert_#{job_prefix} -l h_vmem=7G #{account} \
|
5
|
+
java -Xmx6g -jar #{picard_tools}/SamFormatConverter.jar I=#{sam_file} O=#{bam_file} VALIDATION_STRINGENCY=LENIENT "
|
6
|
+
end
|
7
|
+
|
8
|
+
#rg_and_sorting = "java -jar -Xmx3g ~/Downloads/picard-tools-1.56/picard-tools-1.56/AddOrReplaceReadGroups.jar I=WT_aligned.bam O=WT_aligned_sorted_rg.bam SO=coordinate ID=15 LB=nina_library PL=Illumina PU=ATCATC SM=My_test VALIDATION_STRINGENCY=LENIENT"
|
9
|
+
def self.rg_and_sorting(bam_file, bam_file_sorted, picard_tools, library, index, sample_name, log_file, id, job_prefix, account)
|
10
|
+
cmd = "qsub -o #{log_file} -e #{log_file}_rg_sorting_errors -V -cwd -b y -hold_jid convert_#{job_prefix} -N sort_#{job_prefix} -l h_vmem=7G #{account} \
|
11
|
+
java -Xmx6g -jar #{picard_tools}/AddOrReplaceReadGroups.jar I=#{bam_file} O=#{bam_file_sorted} SO=coordinate ID=#{id} \
|
12
|
+
LB=#{library} PL=Illumina PU=#{index} SM=#{sample_name} VALIDATION_STRINGENCY=LENIENT MAX_RECORDS_IN_RAM=1500000"
|
13
|
+
end
|
14
|
+
|
15
|
+
|
16
|
+
#mark_dublicates = "java -jar ~/Downloads/picard-tools-1.56/picard-tools-1.56/MarkDuplicates.jar I=WT_aligned_sorted_rg.bam O=marked_dublicates.bam M=dublicate.metrcis AS=true VALIDATION_STRINGENCY=LENIENT"
|
17
|
+
def self.mark_dublicates(bam_file_sorted, bam_file_sorted_dublicates, duplicate_metrcis, picard_tools, log_file, job_prefix, account)
|
18
|
+
cmd = "qsub -o #{log_file} -e #{log_file}_duplicates_errors -V -cwd -b y -hold_jid sort_#{job_prefix} -N duplicates_#{job_prefix} -l h_vmem=7G #{account} \
|
19
|
+
java -Xmx3g -jar #{picard_tools}/MarkDuplicates.jar I=#{bam_file_sorted} O=#{bam_file_sorted_dublicates} M=#{duplicate_metrcis} \
|
20
|
+
AS=true VALIDATION_STRINGENCY=LENIENT"
|
21
|
+
end
|
22
|
+
|
23
|
+
#build_index = "java -jar ~/Downloads/picard-tools-1.56/picard-tools-1.56/BuildBamIndex.jar I=marked_dublicates.bam VALIDATION_STRINGENCY=LENIENT"
|
24
|
+
def self.build_index(bam_file_sorted_dublicates, picard_tools, log_file, job_prefix, account)
|
25
|
+
cmd = "qsub -o #{log_file} -e #{log_file}_index_errors -V -cwd -b y -hold_jid dublicates_#{job_prefix} -N index_#{job_prefix} -l h_vmem=7G #{account} \
|
26
|
+
java -Xmx3g -jar #{picard_tools}/BuildBamIndex.jar I=#{bam_file_sorted_dublicates} VALIDATION_STRINGENCY=LENIENT"
|
27
|
+
end
|
28
|
+
end
|
metadata
CHANGED
@@ -1,35 +1,25 @@
|
|
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
|
+
version: 0.0.16
|
5
5
|
prerelease:
|
6
|
-
segments:
|
7
|
-
- 0
|
8
|
-
- 0
|
9
|
-
- 15
|
10
|
-
version: 0.0.15
|
11
6
|
platform: ruby
|
12
|
-
authors:
|
7
|
+
authors:
|
13
8
|
- Kaharina Hayer
|
14
9
|
autorequire:
|
15
10
|
bindir: bin
|
16
11
|
cert_chain: []
|
17
|
-
|
18
|
-
date: 2012-01-20 00:00:00 -05:00
|
19
|
-
default_executable:
|
12
|
+
date: 2012-10-08 00:00:00.000000000 Z
|
20
13
|
dependencies: []
|
21
|
-
|
22
14
|
description: Using BWA to align and GATK to call the bases
|
23
|
-
email:
|
15
|
+
email:
|
24
16
|
- katharinaehayer@gmail.com
|
25
|
-
executables:
|
17
|
+
executables:
|
26
18
|
- mutations_caller_pipeline_aws
|
27
19
|
- gatk_pipe_only_aws
|
28
20
|
extensions: []
|
29
|
-
|
30
21
|
extra_rdoc_files: []
|
31
|
-
|
32
|
-
files:
|
22
|
+
files:
|
33
23
|
- bin/gatk_pipe_only_aws
|
34
24
|
- bin/mutations_caller_pipeline_aws
|
35
25
|
- lib/mutations_caller_pipeline_aws.rb
|
@@ -37,40 +27,30 @@ files:
|
|
37
27
|
- lib/mutations_caller_pipeline_aws/bwa_caller.rb
|
38
28
|
- lib/mutations_caller_pipeline_aws/gatk_caller.rb
|
39
29
|
- lib/mutations_caller_pipeline_aws/location_file.rb
|
30
|
+
- lib/mutations_caller_pipeline_aws/picard_caller.rb
|
40
31
|
- lib/mutations_caller_pipeline_aws/samtools_indexing.rb
|
41
|
-
has_rdoc: true
|
42
32
|
homepage: https://github.com/khayer/mutations_caller_pipeline_aws
|
43
33
|
licenses: []
|
44
|
-
|
45
34
|
post_install_message:
|
46
35
|
rdoc_options: []
|
47
|
-
|
48
|
-
require_paths:
|
36
|
+
require_paths:
|
49
37
|
- lib
|
50
|
-
required_ruby_version: !ruby/object:Gem::Requirement
|
38
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
51
39
|
none: false
|
52
|
-
requirements:
|
53
|
-
- -
|
54
|
-
- !ruby/object:Gem::Version
|
55
|
-
|
56
|
-
|
57
|
-
- 0
|
58
|
-
version: "0"
|
59
|
-
required_rubygems_version: !ruby/object:Gem::Requirement
|
40
|
+
requirements:
|
41
|
+
- - ! '>='
|
42
|
+
- !ruby/object:Gem::Version
|
43
|
+
version: '0'
|
44
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
60
45
|
none: false
|
61
|
-
requirements:
|
62
|
-
- -
|
63
|
-
- !ruby/object:Gem::Version
|
64
|
-
|
65
|
-
segments:
|
66
|
-
- 0
|
67
|
-
version: "0"
|
46
|
+
requirements:
|
47
|
+
- - ! '>='
|
48
|
+
- !ruby/object:Gem::Version
|
49
|
+
version: '0'
|
68
50
|
requirements: []
|
69
|
-
|
70
51
|
rubyforge_project: mutations_caller_pipeline_aws
|
71
|
-
rubygems_version: 1.
|
52
|
+
rubygems_version: 1.8.23
|
72
53
|
signing_key:
|
73
54
|
specification_version: 3
|
74
55
|
summary: Call Mutations for files.fq
|
75
56
|
test_files: []
|
76
|
-
|