mutations_caller_pipeline_aws 0.0.6

data/bin/gatk_pipe_only_aws

@@ -0,0 +1,168 @@
+#!/usr/bin/env ruby
+require 'mutations_caller_pipeline_aws'
+require 'optparse'
+require 'rubygems'
+require 'yaml'
+
+usage =<<EOF
+_________________________________________________________________________________________________
+
+  #{$0}
+        -b sorted_bam_file
+        -c config.yml -v raw_vcf_file
+        [-a account || -p project]
+
+_________________________________________________________________________________________________
+
+  #{$0} ...
+  ... is a tool to find mutations between the reference gene and a given test
+  population. To run this tool you must have bwa, samtools and GATK installed.
+  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  +++
+
+  config.yml should look like this:
+  # config.yml
+  index_prefix: "path/to/prefix"
+  annotation_file: "path/to/annotation_file"
+  bwa: "path/to/bwa"
+  samtools: "path/to/samtools"
+  gatk: "path/to/GenomeAnalysisTK.jar"
+_________________________________________________________________________________________________
+
+EOF
+
+options = { :bam_file_sorted => nil,
+            :index_prefix => nil,
+            :index_fa => nil,
+            :index_vcf => nil,
+            :annotation_file => nil,
+            :samtools => nil,
+            :gatk => nil,
+            :bwa => nil,
+            :vcf => nil,
+            :account => "",
+            :project => "",
+            :debug = 1,
+          }
+
+optparse = OptionParser.new do |opts|
+  opts.banner = usage
+
+  opts.on("-b", "--bam_file_sorted DIR", :REQUIRED, String, "Bam file sorted") do |i|
+    options[:bam_file_sorted] = i
+  end
+
+  opts.on("-c", "--config DIR",:REQUIRED, String, "Set config file") do |path|
+    options.merge!(Hash[YAML::load(open(path)).map { |k, v| [k.to_sym, v] }])
+  end
+
+  opts.on("-v","--vcf [PATH]",:REQUIRED, String, "Output of pipeline") do |i|
+    options[:vcf] = i
+  end
+
+  opts.on("-a","--account [NAME]", "Option for qsub: -A [NAME]. Default: [none] " ) do |i|
+    options[:account] = " -A #{i}" if i
+  end
+
+  opts.on("-p","--project [NAME]", "Option for qsub: -P [NAME]. Default: [none] " ) do |i|
+    options[:project] = " -P #{i}" if i
+  end
+
+  opts.on("-d","--debug ", "Option to debug" ) do |i|
+    options[:debug] =  5 if i
+  end
+
+  opts.on_tail("-h", "--help", "Show this message") do
+    puts opts
+    exit
+  end
+end
+
+begin
+  optparse.parse!
+  mandatory = [: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")}"
+    puts optparse
+    exit
+  end
+rescue OptionParser::InvalidOption, OptionParser::MissingArgument
+   puts $!.to_s
+   puts optparse
+   exit
+end
+
+# pipeline starts here
+
+# tmp files for output
+random = (rand*1000000).floor.to_s
+bam_file = options[:bam_file_sorted]
+job_prefix = "#{random}"
+log_file = "#{random}.log"
+target_intervals = "#{random}_target.intervals"
+realigned_bam = "#{random}_realigned.bam"
+recal_file = "#{random}_recal.csv"
+recal_bam = "#{random}_recal.bam"
+
+options[:account] = options[:project] if options[:account].empty?
+
+
+# Indexing
+SamtoolsIndexing.call(bam_file,
+                      job_prefix,
+                      account,
+                      options[:debug])
+
+# Realigne
+GatkCaller.prepare_realigne(log_file,
+                            options[:gatk],
+                            bam_file,
+                            options[:index_fa],
+                            target_intervals,
+                            job_prefix,
+                            options[:account],
+                            options[:debug])
+
+GatkCaller.realigne(log_file,
+                    options[:gatk],
+                    bam_file,
+                    options[:index_fa],
+                    target_intervals,
+                    realigned_bam,
+                    job_prefix,
+                    options[:account],
+                    options[:debug])
+
+# Recalibration
+GatkCaller.recalibrate_bam( log_file,
+                            options[:gatk],
+                            options[:index_fa],
+                            options[:index_vcf],
+                            realigned_bam,
+                            recal_file,
+                            job_prefix,
+                            options[:account],
+                            options[:debug] )
+
+GatkCaller.table_calibration(log_file,
+                             options[:gatk],
+                             options[:index_fa],
+                             realigned_bam,
+                             recal_bam,
+                             recal_file,
+                             job_prefix,
+                             options[:account],
+                             options[:debug])
+
+# GATK: finding mutations
+GatkCaller.call(log_file,
+                options[:gatk],
+                options[:index_fa],
+                recal_bam,
+                options[:vcf],
+                job_prefix,
+                options[:account],
+                options[:debug])

data/bin/mutations_caller_pipeline_aws

@@ -0,0 +1,196 @@
+#!/usr/bin/env ruby
+require 'mutations_caller_pipeline_aws'
+require 'optparse'
+require 'rubygems'
+require 'yaml'
+
+usage =<<EOF
+_________________________________________________________________________________________________
+
+  #{$0}
+        -m mutant_r1.fq -n mutant_r2.fq
+        [-w wildtype_r1.fq -x wildtype_r2.fq]
+        -c config.yml -v raw_vcf_file
+        [-a account || -p project]
+_________________________________________________________________________________________________
+
+  #{$0} ...
+  ... is a tool to find mutations between the reference gene and a given test
+  population. To run this tool you must have bwa, samtools and GATK installed.
+  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  +++
+
+  config.yml should look like this:
+  # config.yml
+  index_prefix: "path/to/prefix"
+  annotation_file: "path/to/annotation_file"
+  bwa: "path/to/bwa"
+  samtools: "path/to/samtools"
+  gatk: "path/to/GenomeAnalysisTK.jar"
+_________________________________________________________________________________________________
+
+EOF
+
+options = { :mutant_r1 => nil,
+            :mutant_r2 => nil,
+            :wildtype_r1 => nil,
+            :wildtype_r2 => nil,
+            :index_prefix => nil,
+            :index_fa => nil,
+            :index_vcf => nil,
+            :annotation_file => nil,
+            :samtools => nil,
+            :gatk => nil,
+            :bwa => nil,
+            :vcf => nil,
+            :account => "",
+            :project => "",
+            :debug => 1,
+          }
+
+optparse = OptionParser.new do |opts|
+  opts.banner = usage
+
+  opts.on("-m", "--fwd_read_mutant DIR", :REQUIRED, String, "Path to fwd read of mutant") do |i|
+    options[:mutant_r1] = i
+  end
+
+  opts.on("-n", "--rev_read_mutant DIR", :REQUIRED, String, "Path to rev read of mutant") do |i|
+    options[:mutant_r2] = i
+  end
+
+  opts.on("-w", "--fwd_read_wildtype DIR", String, "Path to fwd read of wildtype, not mandatory") do |i|
+    options[:wildtype_r1] = i if i
+  end
+
+  opts.on("-x", "--rev_read_wildtype DIR", String, "Path to rev read of wildtype, not mandatory") do |i|
+    options[:wildtype_r2] = i if i
+  end
+
+  opts.on("-c", "--config DIR", String, "Set config file") do |path|
+    options.merge!(Hash[YAML::load(open(path)).map { |k, v| [k.to_sym, v] }])
+  end
+
+  opts.on("-v","--vcf [PATH]", "Output of pipeline") do |i|
+    options[:vcf] = i
+  end
+
+  opts.on("-a","--account [NAME]", "Option for qsub: -A [NAME]. Default: [none] " ) do |i|
+    options[:account] = " -A #{i}" if i
+  end
+
+  opts.on("-p","--project [NAME]", "Option for qsub: -P [NAME]. Default: [none] " ) do |i|
+    options[:project] = " -P #{i}" if i
+  end
+
+  opts.on("-d","--debug ", "Option to debug" ) do |i|
+    options[:debug] =  5 if i
+  end
+
+  opts.on_tail("-h", "--help", "Show this message") do
+    puts opts
+    exit
+  end
+end
+
+begin
+  optparse.parse!
+  mandatory = [:mutant_r1, :mutant_r2, :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")}"
+    puts optparse
+    exit
+  end
+rescue OptionParser::InvalidOption, OptionParser::MissingArgument
+   puts $!.to_s
+   puts optparse
+   exit
+end
+
+# pipeline starts here
+
+# tmp files for output
+random = (rand*1000000).floor.to_s
+bam_file = "mutant_#{random}"
+job_prefix = "#{random}"
+log_file = "#{random}.log"
+target_intervals = "#{random}_target.intervals"
+realigned_bam = "#{random}_realigned.bam"
+recal_file = "#{random}_recal.csv"
+recal_bam = "#{random}_recal.bam"
+
+options[:account] = options[:project] if options[:account].empty?
+
+# BWA : First step mapping reads to reference
+BwaCaller.call_paired_end(options[:mutant_r1],
+                          options[:mutant_r2],
+                          bam_file,
+                          options[:index_prefix],
+                          log_file,
+                          options[:bwa],
+                          options[:samtools],
+                          job_prefix,
+                          options[:account],
+                          options[:debug])
+
+
+# Indexing
+bam_file = bam_file + ".bam"
+SamtoolsIndexing.call(bam_file,
+                      job_prefix,
+                      options[:account],
+                      options[:debug])
+
+# Realigne
+GatkCaller.prepare_realigne(log_file,
+                            options[:gatk],
+                            bam_file,
+                            options[:index_fa],
+                            target_intervals,
+                            job_prefix,
+                            options[:account],
+                            options[:debug])
+
+GatkCaller.realigne(log_file,
+                    options[:gatk],
+                    bam_file,
+                    options[:index_fa],
+                    target_intervals,
+                    realigned_bam,
+                    job_prefix,
+                    options[:account],
+                    options[:debug])
+
+# Recalibration
+GatkCaller.recalibrate_bam( log_file,
+                            options[:gatk],
+                            options[:index_fa],
+                            options[:index_vcf],
+                            realigned_bam,
+                            recal_file,
+                            job_prefix,
+                            options[:account],
+                            options[:debug] )
+
+GatkCaller.table_calibration(log_file,
+                             options[:gatk],
+                             options[:index_fa],
+                             realigned_bam,
+                             recal_bam,
+                             recal_file,
+                             job_prefix,
+                             options[:account],
+                             options[:debug])
+
+# GATK: finding mutations
+GatkCaller.call(log_file,
+                options[:gatk],
+                options[:index_fa],
+                recal_bam,
+                options[:vcf],
+                job_prefix,
+                options[:account],
+                options[:debug])

data/lib/mutations_caller_pipeline_aws/bwa_caller.rb

@@ -0,0 +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}"
+    puts cmd
+    system('bash','-c',cmd )
+  end
+
+  def self.call_paired_end(r1, r2, out_file, index, log_file, bwa, samtools, job_prefix,account, debug)
+    cmd = "qsub -cwd -b y -N #{job_prefix}_bwa -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} || exit 1) <(#{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}"
+    puts cmd
+    system('bash','-c', cmd) if debug == 1
+  end
+end

data/lib/mutations_caller_pipeline_aws/gatk_caller.rb

@@ -0,0 +1,70 @@
+class GatkCaller
+  # INDEX is normal genom.fa
+  # 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 -cwd -b y -N #{job_prefix}_genotyper -l h_vmem=3G -hold_jid #{job_prefix}_recalibration #{account}\
+      java -Xmx4g -jar  #{gatk} -l INFO -R #{index_fa} -T UnifiedGenotyper \
+      -I #{read_bam} \
+      -o #{read_vcf} \
+      --genotype_likelihoods_model BOTH \
+      >> #{log_dir} 2>&1 || exit 1"
+    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 -cwd -b y -N #{job_prefix}_recalibration_table -l h_vmem=3G  #{account} \
+      java -Xmx4g -jar #{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 "
+    puts cmd
+    system(cmd) if debug == 1
+  end
+
+  # 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 -cwd -b y -N #{job_prefix}_recalibration -l h_vmem=3G -hold_jid #{job_prefix}_recalibration_table #{account} \
+      java -Xmx4g -jar #{gatk} \
+      -R #{index_fa} \
+      -I #{read_bam} \
+      -T TableRecalibration \
+      -o #{recal_bam} \
+      -recalFile #{recal_file} >> #{log_dir} 2>&1 || exit 1"
+    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 -cwd -b y -N #{job_prefix}_prep_realignment -l h_vmem=3G -hold_jid #{job_prefix}_indexing #{account}\
+      java -Xmx2g -jar #{gatk} \
+      -I #{read_bam} \
+      -R #{index_fa} \
+      -T RealignerTargetCreator \
+      -o #{target_intervals}"
+    puts cmd
+    system(cmd) if debug == 1
+  end
+
+  # 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 -cwd -b y -N #{job_prefix}_realignment -l h_vmem=3G -hold_jid #{job_prefix}_prep_realignment #{account} \
+      java -Xmx4g -jar #{gatk} \
+      -I #{read_bam} \
+      -R #{index_fa} \
+      -T IndelRealigner \
+      -targetIntervals #{target_intervals} \
+      -o #{realigned_bam} >> #{log_dir} 2>&1 || exit 1"
+    puts cmd
+    system(cmd) if debug == 1
+  end
+
+end

data/lib/mutations_caller_pipeline_aws/location_file.rb

@@ -0,0 +1,21 @@
+class LocationFile
+  def self.create(vcf_file, loction_file_output)
+    locations = File.open(vcf_file)
+    line = locations.readline()
+
+    locus = []
+    while line.include?('#')
+      location = line.scan(/##contig=<ID=+\w+/)
+      if !location.empty?()
+        location = location[0].split('=')
+        locus << location[-1]
+      end
+      line = locations.readline()
+    end
+
+    locations.close()
+    locus_file = File.new(location_file_output,'w')
+    locus_file.write(locus.join("\n"))
+    locus_file.close()
+  end
+end

data/lib/mutations_caller_pipeline_aws/samtools_indexing.rb

@@ -0,0 +1,8 @@
+class SamtoolsIndexing
+  def self.call(bam_file, job_prefix, account, debug)
+    cmd = "qsub -cwd -b y -N #{job_prefix}_indexing -l h_vmem=3G -hold_jid #{job_prefix}_bwa #{account} \
+      samtools index #{bam_file}"
+    puts cmd
+    system(cmd) if debug == 1
+  end
+end

data/lib/mutations_caller_pipeline_aws.rb

@@ -0,0 +1,11 @@
+require 'mutations_caller_pipeline_aws/bwa_caller'
+require 'mutations_caller_pipeline_aws/gatk_caller'
+require 'mutations_caller_pipeline_aws/samtools_indexing'
+
+class MutationsCallerPipelineAws
+  def self.hi
+    "Hello World!"
+  end
+end
+
+

metadata

@@ -0,0 +1,56 @@
+--- !ruby/object:Gem::Specification
+name: mutations_caller_pipeline_aws
+version: !ruby/object:Gem::Version
+  version: 0.0.6
+  prerelease: 
+platform: ruby
+authors:
+- Kaharina Hayer
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2012-01-20 00:00:00.000000000 Z
+dependencies: []
+description: Using BWA to align and GATK to call the bases
+email:
+- katharinaehayer@gmail.com
+executables:
+- mutations_caller_pipeline_aws
+- gatk_pipe_only_aws
+extensions: []
+extra_rdoc_files: []
+files:
+- bin/gatk_pipe_only_aws
+- bin/mutations_caller_pipeline_aws
+- lib/mutations_caller_pipeline_aws.rb
+- lib/mutations_caller_pipeline_aws/.DS_Store
+- 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/samtools_indexing.rb
+homepage: https://github.com/khayer/mutations_caller_pipeline_aws
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: mutations_caller_pipeline_aws
+rubygems_version: 1.8.10
+signing_key: 
+specification_version: 3
+summary: Call Mutations for files.fq
+test_files: []
+has_rdoc: