mutations_caller_pipeline_aws 0.0.15 → 0.0.16

data/bin/gatk_pipe_only_aws

@@ -4,6 +4,13 @@ require 'optparse'
 require 'rubygems'
 require 'yaml'
 
+###
+#   NOT FUNTIONAL ANYMORE
+###
+
+
+
+
 usage =<<EOF
 _________________________________________________________________________________________________
 
@@ -11,6 +18,7 @@ ________________________________________________________________________________
         -b sorted_bam_file
         -c config.yml -v raw_vcf_file
         [-a account || -p project]
+        -s SampleSheet.csv
 
 _________________________________________________________________________________________________
 
@@ -44,7 +52,8 @@ options = { :bam_file_sorted => nil,
             :vcf => nil,
             :account => "",
             :project => "",
-            :debug => 1
+            :debug => 1,
+            :samplesheet => nil
           }
 
 optparse = OptionParser.new do |opts|
@@ -74,6 +83,10 @@ optparse = OptionParser.new do |opts|
     options[:debug] =  5 if i
   end
 
+  opts.on("-s", "--sampleSheet DIR", :REQUIRED, String, "SampleSheet.csv") do |i|
+    options[:samplesheet] = i
+  end
+
   opts.on_tail("-h", "--help", "Show this message") do
     puts opts
     exit
@@ -82,7 +95,8 @@ end
 
 begin
   optparse.parse!
-  mandatory = [:bam_file_sorted, :index_prefix, :annotation_file, :bwa, :samtools, :gatk, :vcf, :index_vcf, :index_fa]
+  mandatory = [:samplesheet, :bam_file_sorted, :index_prefix, :annotation_file,
+    :bwa, :samtools, :gatk, :vcf, :index_vcf, :index_fa]
   missing = mandatory.select{ |param| options[param].nil? }
   if !missing.empty?
     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?
 
 
 
-# Indexing
+# Indexing + Mark Duplicates
 SamtoolsIndexing.call(bam_file,
                       job_prefix,
                       options[:account],

data/bin/mutations_caller_pipeline_aws

@@ -3,6 +3,14 @@ require 'mutations_caller_pipeline_aws'
 require 'optparse'
 require 'rubygems'
 require 'yaml'
+require 'csv'
+=begin
+  * Name: Mutations Caller Pipeline (AWS)
+  * Pipeline combining bwa with GATK2
+  * Author: Katharina Hayer
+  * Date: 8/8/2012
+  * License: GNU General Public License (GPL-2.0)
+=end
 
 usage =<<EOF
 _________________________________________________________________________________________________
@@ -12,6 +20,8 @@ ________________________________________________________________________________
         [-w wildtype_r1.fq -x wildtype_r2.fq]
         -c config.yml -v raw_vcf_file
         [-a account || -p project]
+        -s SampleSheet.csv
+        -f dbsnp_file
 _________________________________________________________________________________________________
 
   #{$0} ...
@@ -20,12 +30,12 @@ ________________________________________________________________________________
   Also you should have the indices for bwa and GATK prepared.
   NOTE: Only paired end reads are supported!
 
-                      +++  C L U S T E R  V E R S I O N - 0.0.12 +++
+                      +++  C L U S T E R  V E R S I O N - 0.0.15 +++
 
   config.yml should look like this:
   # config.yml
-  index_prefix: "path/to/prefix"
-  annotation_file: "path/to/annotation_file"
+  bwa_prefix: "path/to/prefix"
+  picard_tools: "path/to/picard_tools/"
   bwa: "path/to/bwa"
   samtools: "path/to/samtools"
   gatk: "path/to/GenomeAnalysisTK.jar"
@@ -49,7 +59,8 @@ options = { :mutant_r1 => nil,
             :project => "",
             :debug => 1,
             :cluster => false,
-            :coverage => false
+            :coverage => false,
+            :samplesheet => nil
           }
 
 optparse = OptionParser.new do |opts|
@@ -91,7 +102,7 @@ optparse = OptionParser.new do |opts|
     options[:debug] =  5 if i
   end
 
-  opts.on("-p","--cluster ", "On compute cluster (none AWS)" ) do |i|
+  opts.on("-k","--cluster ", "On compute cluster (none AWS)" ) do |i|
     options[:cluster] =  true if i
   end
 
@@ -99,6 +110,19 @@ optparse = OptionParser.new do |opts|
     options[:coverage] =  true if i
   end
 
+  opts.on("-s", "--sampleSheet DIR", :REQUIRED, String, "SampleSheet.csv") do |i|
+    options[:samplesheet] = i
+  end
+
+  opts.on("-f", "--dbsnp_file DIR", :REQUIRED, String, "dbsnp.vcf") do |i|
+    options[:dbsnp_file] = i
+  end
+
+  opts.on("-b", "--sample_name Name", :REQUIRED, String, "Sample name / SampleID") do |i|
+    options[:sample_name] = i
+  end
+
+
   opts.on_tail("-h", "--help", "Show this message") do
     puts opts
     exit
@@ -107,7 +131,8 @@ end
 
 begin
   optparse.parse!
-  mandatory = [:mutant_r1, :mutant_r2, :index_prefix, :annotation_file, :bwa, :samtools, :gatk, :vcf, :index_vcf, :index_fa]
+  mandatory = [:dbsnp_file,:picard_tools, :samplesheet, :mutant_r1, :mutant_r2, :bwa_prefix,
+    :bwa, :samtools, :gatk, :vcf, :index_vcf, :sample_name, :index_fa]
   missing = mandatory.select{ |param| options[param].nil? }
   if !missing.empty?
     puts "\nMissing options given or missing in config_file: \n\t#{missing.join(",\n\t")}"
@@ -120,41 +145,117 @@ rescue OptionParser::InvalidOption, OptionParser::MissingArgument
    exit
 end
 
+# Methods
+def execute(cmd)
+  puts cmd
+  status = system(cmd)
+  raise cmd unless status
+end
+
+def run(options)
+  # tmp files for output
+  job_number = options[:job_number]
+  bam_file = "aligned_#{job_number}.bam"
+  bam_file_sorted = "sorted_#{job_number}.bam"
+  bam_file_sorted_dublicates = options[:step_one_bam_file]
+  dublicate_metrcis = "GATK_files/dublicate.metrics"
+  job_prefix = "#{job_number}"
+  log_file = "log/#{options[:sample_name]}.log"
+
+  cmd = PicardCaller.convert(options[:sam_file], bam_file, options[:picard_tools],
+    log_file, job_prefix, options[:account])
+
+  if options[:debug] == 1
+    execute(cmd)
+  else
+    puts cmd
+  end
+
+  cmd = PicardCaller.rg_and_sorting(bam_file, bam_file_sorted, options[:picard_tools],
+    options[:library], options[:index], options[:sample_name],
+    log_file, options[:id], job_prefix, options[:account])
+
+  if options[:debug] == 1
+    execute(cmd)
+  else
+    puts cmd
+  end
+
+  cmd = PicardCaller.mark_dublicates(bam_file_sorted, bam_file_sorted_dublicates,
+    dublicate_metrcis, options[:picard_tools], log_file, job_prefix,
+    options[:account])
+
+  if options[:debug] == 1
+    execute(cmd)
+  else
+    puts cmd
+  end
+
+  cmd = PicardCaller.build_index(bam_file_sorted_dublicates,
+    options[:picard_tools], log_file, job_prefix, options[:account])
+
+  if options[:debug] == 1
+    execute(cmd)
+  else
+    puts cmd
+  end
+end
+
+
 # pipeline starts here
 
+# get information from sample sheet
+CSV.foreach(options[:samplesheet],{:headers => :first_row}) do |row|
+  if sample_name = row["SampleID"]
+    options[:index] = row["Index"]
+    lane = row["Lane"]
+    sample_project = row["SampleProject"]
+    options[:id] = "#{sample_project}_#{lane}_#{sample_name}"
+    options[:library] = row["FCID"]
+  end
+end
+
+# Create Dir to not overcluster output folder
+Dir.mkdir("GATK_files") unless File.exists?("GATK_files")
+Dir.mkdir("log") unless File.exists?("log")
+
 # tmp files for output
 random = (rand*1000000).floor.to_s
-bam_file = "mutant_#{random}"
+sample_name = options[:sample_name]
+options[:sam_file] = "#{sample_name}_#{random}.sam"
+options[:step_one_bam_file] = "#{sample_name}_#{random}.bam"
+bam_file = "mutant_#{random}.bam"
 job_prefix = "#{random}"
-log_file = "#{random}.log"
-target_intervals = "#{random}_target.intervals"
+options[:job_number] = job_prefix
+log_file = "log/#{sample_name}.log"
+target_intervals = "GATK_files/target.intervals"
 realigned_bam = "#{random}_realigned.bam"
-recal_file = "#{random}_recal.csv"
-recal_bam = "#{random}_recal.bam"
+recal_file = "GATK_files/recal.grp"
+recal_bam = "#{sample_name}.bam"
 sai_file_fwd = "#{random}_fwd.sai"
 sai_file_rev = "#{random}_rev.sai"
 
 options[:account] = options[:project] if options[:account].empty?
-options[:gatk] = "java -Xmx4g -jar #{options[:gatk]}" if options[:cluster]
+#options[:gatk] = "java -Xmx8g -jar #{options[:gatk]}"
 
 # BWA : ALN
 BwaCaller.call_aln(options[:mutant_r1],
-                   options[:index_prefix],
+                   options[:bwa_prefix],
                    sai_file_fwd,
                    log_file,
                    options[:bwa],
                    job_prefix,
                    options[:account],
-                   options[:debug])
+                   options[:debug],"fwd")
 
 BwaCaller.call_aln(options[:mutant_r2],
-                   options[:index_prefix],
+                   options[:bwa_prefix],
                    sai_file_rev,
                    log_file,
                    options[:bwa],
                    job_prefix,
                    options[:account],
-                   options[:debug])
+                   options[:debug],"rev")
 
 
 # BWA : First step mapping reads to reference
@@ -162,8 +263,8 @@ BwaCaller.call_paired_end(options[:mutant_r1],
                           options[:mutant_r2],
                           sai_file_fwd,
                           sai_file_rev,
-                          bam_file,
-                          options[:index_prefix],
+                          options[:sam_file],
+                          options[:bwa_prefix],
                           log_file,
                           options[:bwa],
                           options[:samtools],
@@ -172,27 +273,30 @@ BwaCaller.call_paired_end(options[:mutant_r1],
                           options[:debug])
 
 
-# Indexing
-bam_file = bam_file + ".bam"
-SamtoolsIndexing.call(bam_file,
-                      job_prefix,
-                      options[:account],
-                      options[:debug],
-                      log_file)
+# Indexing + Marking Duplicates
+run(options)
+#bam_file = bam_file + ".bam"
+#SamtoolsIndexing.call(bam_file,
+#                      job_prefix,
+#                      options[:account],
+#                      options[:debug],
+#                      log_file)
 
 # Realigne
 GatkCaller.prepare_realigne(log_file,
                             options[:gatk],
-                            bam_file,
+                            options[:step_one_bam_file],
                             options[:index_fa],
                             target_intervals,
                             job_prefix,
                             options[:account],
+                            options[:dbsnp_file],
                             options[:debug])
 
+
 GatkCaller.realigne(log_file,
                     options[:gatk],
-                    bam_file,
+                    options[:step_one_bam_file],
                     options[:index_fa],
                     target_intervals,
                     realigned_bam,
@@ -204,11 +308,11 @@ GatkCaller.realigne(log_file,
 GatkCaller.recalibrate_bam( log_file,
                             options[:gatk],
                             options[:index_fa],
-                            options[:index_vcf],
                             realigned_bam,
                             recal_file,
                             job_prefix,
                             options[:account],
+                            options[:dbsnp_file],
                             options[:debug] )
 
 GatkCaller.table_calibration(log_file,
@@ -225,11 +329,11 @@ GatkCaller.table_calibration(log_file,
 GatkCaller.call(log_file,
                 options[:gatk],
                 options[:index_fa],
-                options[:index_vcf],
                 recal_bam,
                 options[:vcf],
                 job_prefix,
                 options[:account],
+                options[:dbsnp_file],
                 options[:debug])
 
 

data/lib/mutations_caller_pipeline_aws.rb

@@ -1,6 +1,7 @@
 require 'mutations_caller_pipeline_aws/bwa_caller'
 require 'mutations_caller_pipeline_aws/gatk_caller'
 require 'mutations_caller_pipeline_aws/samtools_indexing'
+require 'mutations_caller_pipeline_aws/picard_caller'
 
 class MutationsCallerPipelineAws
   def self.hi

data/lib/mutations_caller_pipeline_aws/bwa_caller.rb

@@ -1,18 +1,17 @@
 class BwaCaller
 
-  def self.call_paired_end(r1, r2, sai1, sai2,  out_file, index, log_file, bwa, samtools, job_prefix,account, debug)
+  def self.call_paired_end(r1, r2, sai1, sai2,  sam_file, index, log_file, bwa, samtools, job_prefix,account, debug)
     dummy = "\\\\\\"
-    cmd = "qsub -o #{log_file} -hold_jid bwa_aln_#{job_prefix} -V -cwd -b y -N bwa_#{job_prefix} -l h_vmem=6G -pe make 3  #{account}\
-           #{bwa} sampe -r '@RG#{dummy}tID:foo#{dummy}tSM:bar#{dummy}tPL:Illumina' #{index} \
-          #{sai1} #{sai2}  #{r1} #{r2} \
-          \\| #{samtools} view -Su - \\| #{samtools} sort - #{out_file}"
+    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}\
+           #{bwa} sampe  #{index}   \
+          #{sai1} #{sai2}  #{r1} #{r2} -f #{sam_file}"
     puts cmd
     system('bash','-c', cmd) if debug == 1
   end
 
-  def self.call_aln(read, index, out_file, log_file, bwa, job_prefix, account,debug)
-    cmd = "qsub -o #{log_file} -V -cwd -b y -N bwa_aln_#{job_prefix} -l h_vmem=4G #{account} \
-           #{bwa} aln -f #{out_file} #{index} #{read} "
+  def self.call_aln(read, index, out_file, log_file, bwa, job_prefix, account,debug,direction)
+    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} \
+           #{bwa} aln -t 8 -f #{out_file} #{index} #{read} "
     puts cmd
     system(cmd) if debug == 1
   end

data/lib/mutations_caller_pipeline_aws/gatk_caller.rb

@@ -1,13 +1,11 @@
 class GatkCaller
   # INDEX is normal genom.fa
   # Genotyper
-  def self.call(log_dir, gatk, index_fa, dbSNP, read_bam, read_vcf, job_prefix, account, debug)
-    cmd = "echo 'starting GATK for mutant at ' `date` >> #{log_dir}
-      qsub -o #{log_dir} -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=4.5G -hold_jid recalibration_#{job_prefix} #{account}\
-      #{gatk} -l INFO -R #{index_fa} -T UnifiedGenotyper \
-      -I #{read_bam} \
-      -D #{dbSNP}
-      -o #{read_vcf} \
+  def self.call(log_dir, gatk, index_fa, read_bam, read_vcf, job_prefix, account,dbsnp_file, debug)
+    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}\
+      java -Xmx6g -jar #{gatk} -l INFO -R #{index_fa} -T UnifiedGenotyper \
+      -I #{read_bam} --dbsnp #{dbsnp_file} \
+      -o #{read_vcf} -nt 4 --max_alternate_alleles 8 \
       --genotype_likelihoods_model BOTH"
     puts cmd
     system(cmd) if debug == 1
@@ -15,49 +13,46 @@ class GatkCaller
 
   # INDEX is normal genom.fa
   # Coverage Summary
+  # parallel not possible yet (1.6-13-g91f02df)
   def self.coverage(log_dir, gatk, index_fa, read_bam, outfile_prefix, job_prefix, account, debug)
-    cmd = "echo 'starting coverage GATK for mutant at ' `date` >> #{log_dir}
-      qsub -o #{log_dir} -V -cwd -b y -N coverage_#{job_prefix} -l h_vmem=4.5G -hold_jid recalibration_#{job_prefix} #{account}\
-      #{gatk} -R #{index_fa} -T DepthOfCoverage \
-      -I #{read_bam} \
-      -o #{outfile_prefix} "
+    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}\
+      java -Xmx6g -jar #{gatk} -R #{index_fa} -T DepthOfCoverage \
+      -I #{read_bam} --omitDepthOutputAtEachBase \
+      -o #{outfile_prefix} --omitIntervalStatistics --omitLocusTable"
     puts cmd
     system(cmd) if debug == 1
   end
 
   # Making recalibration table
-  def self.recalibrate_bam(log_dir ,gatk, index_fa, index_vcf, read_bam, recal_file, job_prefix, account, debug )
-    cmd = "echo 'starting recalibration table ' `date` >> #{log_dir}
-      qsub -o #{log_dir} -V -cwd -b y -N recalibration_table_#{job_prefix} -l h_vmem=4.5G  -hold_jid realignment_#{job_prefix} #{account} \
-      #{gatk} -knownSites #{index_vcf} -I #{read_bam} \
-      -R #{index_fa} -T CountCovariates \
-      -cov ReadGroupCovariate -cov QualityScoreCovariate -cov DinucCovariate \
-      -cov CycleCovariate \
-      -recalFile #{recal_file}"
+  def self.recalibrate_bam(log_dir ,gatk, index_fa, read_bam, recal_file, job_prefix, account, dbsnp_file, debug )
+    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} \
+      java -Xmx6g -jar #{gatk} -knownSites #{dbsnp_file} -I #{read_bam} \
+      -R #{index_fa} -T BaseRecalibrator  \
+      -o #{recal_file}"
     puts cmd
     system(cmd) if debug == 1
   end
 
   # Using recalibration table
+  # parallel not possible yet (1.6-13-g91f02df)
   def self.table_calibration(log_dir, gatk, index_fa, read_bam, recal_bam, recal_file, job_prefix, account, debug)
-    cmd = "echo 'recalibrating bam_file at ' `date` >> #{log_dir}
-      qsub -V -o #{log_dir} -cwd -b y -N recalibration_#{job_prefix} -l h_vmem=4.5G -hold_jid recalibration_table_#{job_prefix} #{account} \
-      #{gatk} \
+    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} \
+      java -Xmx6g -jar #{gatk} \
       -R #{index_fa} \
       -I #{read_bam} \
-      -T TableRecalibration \
+      -T PrintReads \
       -o #{recal_bam} \
-      -recalFile #{recal_file}"
+      -BQSR #{recal_file}"
     puts cmd
     system(cmd) if debug == 1
   end
 
   # Preparation realignement
-  def self.prepare_realigne(log_dir, gatk, read_bam, index_fa, target_intervals, job_prefix, account, debug)
-    cmd = "echo 'preparing realignement at ' `date` >> #{log_dir}
-      qsub -o #{log_dir} -V -cwd -b y -N prep_realignment_#{job_prefix} -l h_vmem=4.5G -hold_jid indexing_#{job_prefix} #{account}\
-      #{gatk} \
-      -I #{read_bam} \
+
+  def self.prepare_realigne(log_dir, gatk, read_bam, index_fa, target_intervals, job_prefix, account, dbsnp_file, debug)
+    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}\
+      java -Xmx6g -jar #{gatk} -nt 4 \
+      -I #{read_bam} --known #{dbsnp_file} \
       -R #{index_fa} \
       -T RealignerTargetCreator \
       -o #{target_intervals}"
@@ -66,10 +61,10 @@ class GatkCaller
   end
 
   # Realignment
+  # parallel not possible yet (1.6-13-g91f02df)
   def self.realigne(log_dir, gatk, read_bam, index_fa, target_intervals, realigned_bam, job_prefix, account, debug)
-    cmd = "echo 'preparing realignement at ' `date` >> #{log_dir}
-      qsub -o #{log_dir} -V -cwd -b y -N realignment_#{job_prefix} -l h_vmem=4.5G -hold_jid prep_realignment_#{job_prefix} #{account} \
-      #{gatk} \
+    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} \
+      java -Xmx6g -jar #{gatk} \
       -I #{read_bam} \
       -R #{index_fa} \
       -T IndelRealigner \

data/lib/mutations_caller_pipeline_aws/picard_caller.rb

@@ -0,0 +1,28 @@
+class PicardCaller
+  #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"
+  def self.convert(sam_file, bam_file, picard_tools, log_file, job_prefix, account)
+    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} \
+      java -Xmx6g -jar  #{picard_tools}/SamFormatConverter.jar I=#{sam_file} O=#{bam_file} VALIDATION_STRINGENCY=LENIENT "
+  end
+
+  #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"
+  def self.rg_and_sorting(bam_file, bam_file_sorted, picard_tools, library, index, sample_name, log_file, id, job_prefix, account)
+    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} \
+      java -Xmx6g -jar  #{picard_tools}/AddOrReplaceReadGroups.jar I=#{bam_file} O=#{bam_file_sorted} SO=coordinate ID=#{id} \
+      LB=#{library} PL=Illumina PU=#{index} SM=#{sample_name} VALIDATION_STRINGENCY=LENIENT MAX_RECORDS_IN_RAM=1500000"
+  end
+
+
+  #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"
+  def self.mark_dublicates(bam_file_sorted, bam_file_sorted_dublicates, duplicate_metrcis, picard_tools, log_file, job_prefix, account)
+    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} \
+      java -Xmx3g -jar #{picard_tools}/MarkDuplicates.jar I=#{bam_file_sorted} O=#{bam_file_sorted_dublicates} M=#{duplicate_metrcis} \
+      AS=true VALIDATION_STRINGENCY=LENIENT"
+  end
+
+  #build_index = "java -jar ~/Downloads/picard-tools-1.56/picard-tools-1.56/BuildBamIndex.jar I=marked_dublicates.bam  VALIDATION_STRINGENCY=LENIENT"
+  def self.build_index(bam_file_sorted_dublicates, picard_tools, log_file, job_prefix, account)
+    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} \
+      java -Xmx3g -jar #{picard_tools}/BuildBamIndex.jar I=#{bam_file_sorted_dublicates}  VALIDATION_STRINGENCY=LENIENT"
+  end
+end

metadata

@@ -1,35 +1,25 @@
---- !ruby/object:Gem::Specification 
+--- !ruby/object:Gem::Specification
 name: mutations_caller_pipeline_aws
-version: !ruby/object:Gem::Version 
-  hash: 1
+version: !ruby/object:Gem::Version
+  version: 0.0.16
   prerelease: 
-  segments: 
-  - 0
-  - 0
-  - 15
-  version: 0.0.15
 platform: ruby
-authors: 
+authors:
 - Kaharina Hayer
 autorequire: 
 bindir: bin
 cert_chain: []
-
-date: 2012-01-20 00:00:00 -05:00
-default_executable: 
+date: 2012-10-08 00:00:00.000000000 Z
 dependencies: []
-
 description: Using BWA to align and GATK to call the bases
-email: 
+email:
 - katharinaehayer@gmail.com
-executables: 
+executables:
 - mutations_caller_pipeline_aws
 - gatk_pipe_only_aws
 extensions: []
-
 extra_rdoc_files: []
-
-files: 
+files:
 - bin/gatk_pipe_only_aws
 - bin/mutations_caller_pipeline_aws
 - lib/mutations_caller_pipeline_aws.rb
@@ -37,40 +27,30 @@ files:
 - lib/mutations_caller_pipeline_aws/bwa_caller.rb
 - lib/mutations_caller_pipeline_aws/gatk_caller.rb
 - lib/mutations_caller_pipeline_aws/location_file.rb
+- lib/mutations_caller_pipeline_aws/picard_caller.rb
 - lib/mutations_caller_pipeline_aws/samtools_indexing.rb
-has_rdoc: true
 homepage: https://github.com/khayer/mutations_caller_pipeline_aws
 licenses: []
-
 post_install_message: 
 rdoc_options: []
-
-require_paths: 
+require_paths:
 - lib
-required_ruby_version: !ruby/object:Gem::Requirement 
+required_ruby_version: !ruby/object:Gem::Requirement
   none: false
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      hash: 3
-      segments: 
-      - 0
-      version: "0"
-required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      hash: 3
-      segments: 
-      - 0
-      version: "0"
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
 requirements: []
-
 rubyforge_project: mutations_caller_pipeline_aws
-rubygems_version: 1.6.2
+rubygems_version: 1.8.23
 signing_key: 
 specification_version: 3
 summary: Call Mutations for files.fq
 test_files: []
-