mutations_caller_pipeline_aws 0.0.10 → 0.0.11

data/bin/gatk_pipe_only_aws

@@ -107,6 +107,7 @@ realigned_bam = "#{random}_realigned.bam"
 recal_file = "#{random}_recal.csv"
 recal_bam = "#{random}_recal.bam"
 
+
 options[:account] = options[:project] if options[:account].empty?
 
 
@@ -114,7 +115,8 @@ options[:account] = options[:project] if options[:account].empty?
 SamtoolsIndexing.call(bam_file,
                       job_prefix,
                       options[:account],
-                      options[:debug])
+                      options[:debug],
+                      log_file)
 
 # Realigne
 GatkCaller.prepare_realigne(log_file,

data/bin/mutations_caller_pipeline_aws

@@ -121,12 +121,36 @@ target_intervals = "#{random}_target.intervals"
 realigned_bam = "#{random}_realigned.bam"
 recal_file = "#{random}_recal.csv"
 recal_bam = "#{random}_recal.bam"
+sai_file_fwd = "#{random}_fwd.sai"
+sai_file_rev = "#{random}_rev.sai"
 
 options[:account] = options[:project] if options[:account].empty?
 
+# BWA : ALN
+BwaCaller.call_aln(options[:mutant_r1],
+                   options[:index_prefix],
+                   sai_file_fwd,
+                   log_file,
+                   options[:bwa],
+                   job_prefix,
+                   options[:account],
+                   options[:debug])
+
+BwaCaller.call_aln(options[:mutant_r2],
+                   options[:index_prefix],
+                   sai_file_rev,
+                   log_file,
+                   options[:bwa],
+                   job_prefix,
+                   options[:account],
+                   options[:debug])
+
+
 # BWA : First step mapping reads to reference
 BwaCaller.call_paired_end(options[:mutant_r1],
                           options[:mutant_r2],
+                          sai_file_fwd,
+                          sai_file_rev,
                           bam_file,
                           options[:index_prefix],
                           log_file,
@@ -142,7 +166,8 @@ bam_file = bam_file + ".bam"
 SamtoolsIndexing.call(bam_file,
                       job_prefix,
                       options[:account],
-                      options[:debug])
+                      options[:debug],
+                      log_file)
 
 # Realigne
 GatkCaller.prepare_realigne(log_file,

data/lib/mutations_caller_pipeline_aws/bwa_caller.rb

@@ -1,18 +1,18 @@
 class BwaCaller
-  def self.call_single_end(r1,out_file,index, log_file, bwa, samtools)
-    cmd = "#{bwa} samse -r '@RG\tID:foo\tSM:bar\tPL:Illumina' #{index} \
-        <(#{bwa} aln #{index} #{r1} 2>>#{log_file})  \
-        #{r1} 2>>#{log_file} | #{samtools} view -Su - 2>>#{log_file} | #{samtools} sort - #{out_file} 2>>#{log_file}"
+
+  def self.call_paired_end(r1, r2, sai1, sai2,  out_file, index, log_file, bwa, samtools, job_prefix,account, debug)
+    cmd = "qsub -o #{log_file} -hold_jid bwa_aln_#{job_prefix} -V -cwd -b y -N bwa_#{job_prefix} -l h_vmem=9G -pe make 3  #{account}\
+           #{bwa} sampe -r '@RG\tID:foo\tSM:bar\tPL:Illumina' #{index} \
+          #{sai1} #{sai2}  #{r1} #{r2} \
+          \\| #{samtools} view -Su - \\| #{samtools} sort - #{out_file}"
     puts cmd
-    system('bash','-c',cmd )
+    system('bash','-c', cmd) if debug == 1
   end
 
-  def self.call_paired_end(r1, r2, out_file, index, log_file, bwa, samtools, job_prefix,account, debug)
-    cmd = "qsub -V -cwd -b y -N bwa_#{job_prefix} -l h_vmem=9G -pe make 4  #{account}\
-      #{bwa} sampe -r '@RG\tID:foo\tSM:bar\tPL:Illumina' #{index} \
-        <(#{bwa} aln #{index} #{r1} 2>>#{log_file}) <(#{bwa} aln #{index} #{r2} 2>>#{log_file} ) \
-        #{r1} #{r2} 2>>#{log_file} | #{samtools} view -Su - 2>>#{log_file} | #{samtools} sort - #{out_file} 2>>#{log_file}"
+  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} "
     puts cmd
-    system('bash','-c', cmd) if debug == 1
+    system(cmd) if debug == 1
   end
 end

data/lib/mutations_caller_pipeline_aws/gatk_caller.rb

@@ -3,12 +3,11 @@ class GatkCaller
   # Genotyper
   def self.call(log_dir, gatk, index_fa, read_bam, read_vcf, job_prefix, account, debug)
     cmd = "echo 'starting GATK for mutant at ' `date` >> #{log_dir}
-      qsub -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=3G -hold_jid recalibration_#{job_prefix} #{account}\
+      qsub -o #{log_dir} -V -cwd -b y -N genotyper_#{job_prefix} -l h_vmem=3G -hold_jid recalibration_#{job_prefix} #{account}\
       #{gatk} -l INFO -R #{index_fa} -T UnifiedGenotyper \
       -I #{read_bam} \
       -o #{read_vcf} \
-      --genotype_likelihoods_model BOTH \
-      >> #{log_dir} 2>&1 || exit 1"
+      --genotype_likelihoods_model BOTH || exit 1"
     puts cmd
     system(cmd) if debug == 1
   end
@@ -16,12 +15,12 @@ class GatkCaller
   # 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 -V -cwd -b y -N recalibration_table_#{job_prefix} -l h_vmem=4G  -hold_jid realignment_#{job_prefix} #{account} \
+      qsub -o #{log_dir} -V -cwd -b y -N recalibration_table_#{job_prefix} -l h_vmem=4G  -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} >> #{log_dir} 2>&1 || exit 1 "
+      -recalFile #{recal_file}  || exit 1 "
     puts cmd
     system(cmd) if debug == 1
   end
@@ -29,13 +28,13 @@ class GatkCaller
   # Using recalibration table
   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 -cwd -b y -N recalibration_#{job_prefix} -l h_vmem=4G -hold_jid recalibration_table_#{job_prefix} #{account} \
+      qsub -V -o #{log_dir} -cwd -b y -N recalibration_#{job_prefix} -l h_vmem=4G -hold_jid recalibration_table_#{job_prefix} #{account} \
       #{gatk} \
       -R #{index_fa} \
       -I #{read_bam} \
       -T TableRecalibration \
       -o #{recal_bam} \
-      -recalFile #{recal_file} >> #{log_dir} 2>&1 || exit 1"
+      -recalFile #{recal_file} || exit 1"
     puts cmd
     system(cmd) if debug == 1
   end
@@ -43,12 +42,12 @@ class GatkCaller
   # 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 -V -cwd -b y -N prep_realignment_#{job_prefix} -l h_vmem=4G -hold_jid indexing_#{job_prefix} #{account}\
+      qsub -o #{log_dir} -V -cwd -b y -N prep_realignment_#{job_prefix} -l h_vmem=4G -hold_jid indexing_#{job_prefix} #{account}\
       #{gatk} \
       -I #{read_bam} \
       -R #{index_fa} \
       -T RealignerTargetCreator \
-      -o #{target_intervals}"
+      -o #{target_intervals} || exit 1 "
     puts cmd
     system(cmd) if debug == 1
   end
@@ -56,13 +55,13 @@ class GatkCaller
   # Realignment
   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 -V -cwd -b y -N realignment_#{job_prefix} -l h_vmem=4G -hold_jid prep_realignment_#{job_prefix} #{account} \
+      qsub -o #{log_dir} -V -cwd -b y -N realignment_#{job_prefix} -l h_vmem=4G -hold_jid prep_realignment_#{job_prefix} #{account} \
       #{gatk} \
       -I #{read_bam} \
       -R #{index_fa} \
       -T IndelRealigner \
       -targetIntervals #{target_intervals} \
-      -o #{realigned_bam} >> #{log_dir} 2>&1 || exit 1"
+      -o #{realigned_bam}  || exit 1"
     puts cmd
     system(cmd) if debug == 1
   end

data/lib/mutations_caller_pipeline_aws/samtools_indexing.rb

@@ -1,6 +1,6 @@
 class SamtoolsIndexing
-  def self.call(bam_file, job_prefix, account, debug)
-    cmd = "qsub -V -cwd -b y -N indexing_#{job_prefix} -l h_vmem=3G -hold_jid bwa_#{job_prefix} #{account} \
+  def self.call(bam_file, job_prefix, account, debug, log_file)
+    cmd = "qsub -o #{log_file} -V -cwd -b y -N indexing_#{job_prefix} -l h_vmem=3G -hold_jid bwa_#{job_prefix} #{account} \
       samtools index #{bam_file}"
     puts cmd
     system(cmd) if debug == 1

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mutations_caller_pipeline_aws
 version: !ruby/object:Gem::Version
-  version: 0.0.10
+  version: 0.0.11
   prerelease: 
 platform: ruby
 authors: