fastq-factory 0.1.3 → 0.1.4

data/VERSION

@@ -1 +1 @@
-0.1.3
+0.1.4

data/bin/fastq-factory

@@ -15,6 +15,9 @@ options = Trollop::options do
   opt :quality_scale, "The Phred quality scale, 33 or 64", :type => :integer, :required => true
   opt :fastq_quality_trimmer_dir, "The directory where fastq_quality_trimmer can be found", :short => "-F", :type => :string
   opt :quake_dir, "The directory where the quake executable can be found", :short => "-Q", :type => :string
+  opt :trim_point_fraction, "The fraction of the read length below which a read will be discarded if it is shorter than teh value after trimming", :short => "-T", :type => :float, :default => 0.6
+  opt :trim_quality_cutoff, "The quality used as a cutoff with which to trim a read from the 3' end", :short => "-E", :type => :integer, :default => 20
+  opt :metrics_quality_cutoff, "The quality value which will be used to determine at which position the read falls below this valeu in a 5 position window", :short => "-W", :type => :integer, :default => 30
 end
 
 forward_reads_suffix, forward_reads_file_extension = options[:forward_reads_suffix].match(/(.+)\.(.+?)$/).captures
@@ -32,7 +35,7 @@ if options[:trim]
   quake_path = find_executable("quake.py", options[:quake_dir])
   abort ("Can not find quake.py. You can specifiy the directory where this can be found using the -Q option") unless quake_path
   write_out_fastq_trim_script
-  trim_and_correct_fastqs(sample_map, options[:directory], forward_reads_suffix, forward_reads_file_extension, reverse_reads_suffix, reverse_reads_file_extension, options[:quality_scale], fastq_quality_trimmer_path, quake_path)
+  trim_and_correct_fastqs(sample_map, options[:directory], forward_reads_suffix, forward_reads_file_extension, reverse_reads_suffix, reverse_reads_file_extension, options[:quality_scale], fastq_quality_trimmer_path, quake_path, options[:trim_point_fraction], options[:trim_quality_cutoff])
 end
 
 if options[:metrics]
@@ -40,5 +43,5 @@ if options[:metrics]
     file_exists?("#{sample_prefix}#{forward_reads_suffix}.trimmed.cor.#{forward_reads_file_extension}", options[:directory])
     file_exists?("#{sample_prefix}#{reverse_reads_suffix}.trimmed.cor.#{reverse_reads_file_extension}", options[:directory])
   end
-  generate_quality_metrics(sample_map, options[:directory], options[:forward_reads_suffix], options[:reverse_reads_suffix], options[:quality_scale])
+  generate_quality_metrics(sample_map, options[:directory], options[:forward_reads_suffix], options[:reverse_reads_suffix], options[:quality_scale], options[:metrics_quality_cutoff])
 end

data/fastq-factory.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = "fastq-factory"
-  s.version = "0.1.3"
+  s.version = "0.1.4"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Anthony Underwood"]
-  s.date = "2012-08-23"
+  s.date = "2012-08-24"
   s.description = "This tool can process fastq files, using fastq_quality_trimmer and quake to correct fastq files and then provide a quality asssessment of the data"
   s.email = "anthony.underwood@hpa.org.uk"
   s.executables = ["fastq-factory"]

data/lib/generate_quality_metrics.rb

@@ -3,8 +3,9 @@ include FastqAssessment
 require 'miseq_run_stats'
 include MiseqRunStats
 
-def generate_quality_metrics(sample_map, directory, forward_reads_suffix, reverse_reads_suffix, quality_scale)
+def generate_quality_metrics(sample_map, directory, forward_reads_suffix, reverse_reads_suffix, quality_scale, quality_cutoff)
   if File.exists?("#{directory}/ResequencingRunStatistics.xml")
+    puts "Assessing quality from Miseq run stats file"
     resequencing_run_stats = parse_resequencing_run_stats("#{directory}/ResequencingRunStatistics.xml", sample_map.values)
   else
     resequencing_run_stats = ResequencingRunStats.new
@@ -20,13 +21,13 @@ def generate_quality_metrics(sample_map, directory, forward_reads_suffix, revers
 
 
   sample_map.each do |read_file_prefix, sample_name|
-    puts sample_name
+    puts "Assesing quality for #{sample_name}"
     resequencing_run_stats.sample_stats[sample_name].fastq_stats = Hash.new
-    resequencing_run_stats.sample_stats[sample_name].fastq_stats["forward"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{forward_reads_suffix}",quality_scale)
-    resequencing_run_stats.sample_stats[sample_name].fastq_stats["reverse"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{reverse_reads_suffix}",quality_scale)
-    resequencing_run_stats.sample_stats[sample_name].fastq_stats["forward-trim_corrected"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{forward_reads_trimmed_corrected_suffix}",quality_scale)
+    resequencing_run_stats.sample_stats[sample_name].fastq_stats["forward"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{forward_reads_suffix}",quality_scale, quality_cutoff)
+    resequencing_run_stats.sample_stats[sample_name].fastq_stats["reverse"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{reverse_reads_suffix}",quality_scale, quality_cutoff)
+    resequencing_run_stats.sample_stats[sample_name].fastq_stats["forward-trim_corrected"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{forward_reads_trimmed_corrected_suffix}",quality_scale, quality_cutoff)
     resequencing_run_stats.sample_stats[sample_name].fastq_stats["forward-trim_corrected"].percentage_compared_to_raw = percentage_compared_to_raw("#{directory}/#{read_file_prefix}#{forward_reads_trimmed_corrected_suffix}", "#{directory}/#{read_file_prefix}#{forward_reads_suffix}")
-    resequencing_run_stats.sample_stats[sample_name].fastq_stats["reverse-trim_corrected"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{reverse_reads_trimmed_corrected_suffix}",quality_scale)
+    resequencing_run_stats.sample_stats[sample_name].fastq_stats["reverse-trim_corrected"] = generate_quality_stats_for_read("#{directory}/#{read_file_prefix}#{reverse_reads_trimmed_corrected_suffix}",quality_scale, quality_cutoff)
     resequencing_run_stats.sample_stats[sample_name].fastq_stats["reverse-trim_corrected"].percentage_compared_to_raw = percentage_compared_to_raw("#{directory}/#{read_file_prefix}#{reverse_reads_trimmed_corrected_suffix}", "#{directory}/#{read_file_prefix}#{reverse_reads_suffix}")
   end
   # print out data

data/lib/trim_and_correct.rb

@@ -1,25 +1,29 @@
-def trim_and_correct_fastqs(sample_map, directory, forward_reads_suffix, forward_reads_file_extension, reverse_reads_suffix, reverse_reads_file_extension, quality_scale, fastq_quality_trimmer_path, quake_path)
+def trim_and_correct_fastqs(sample_map, directory, forward_reads_suffix, forward_reads_file_extension, reverse_reads_suffix, reverse_reads_file_extension, quality_scale, fastq_quality_trimmer_path, quake_path,trim_point_fraction, trim_quality_cutoff)
   Dir.chdir(directory)
   # trimming
   sample_map.each do |sample_file_prefix, sample_name|
     puts "Trimming files for #{sample_name}"
+    #determine read length
+    read_length = calculate_read_length("#{directory}/#{sample_file_prefix}#{forward_reads_suffix}.#{forward_reads_file_extension}")
+    trim_point = (trim_point_fraction * read_length).to_i
 
-    system("#{fastq_quality_trimmer_path} -i #{directory}/#{sample_file_prefix}#{forward_reads_suffix}.#{forward_reads_file_extension} -o #{directory}/#{sample_file_prefix}#{forward_reads_suffix}.trimmed.#{forward_reads_file_extension} -t 20 -l 90 -Q #{quality_scale} -v")
-    system("#{fastq_quality_trimmer_path} -i #{directory}/#{sample_file_prefix}#{reverse_reads_suffix}.#{reverse_reads_file_extension} -o #{directory}/#{sample_file_prefix}#{reverse_reads_suffix}.trimmed.#{reverse_reads_file_extension} -t 20 -l 90 -Q #{quality_scale} -v")
-    system("perl /tmp/fastq-remove-orphans.pl -1 #{sample_file_prefix}#{forward_reads_suffix}.trimmed.#{forward_reads_file_extension} -2 #{sample_file_prefix}#{reverse_reads_suffix}.trimmed.#{reverse_reads_file_extension}")
+    `#{fastq_quality_trimmer_path} -i #{directory}/#{sample_file_prefix}#{forward_reads_suffix}.#{forward_reads_file_extension} -o #{directory}/#{sample_file_prefix}#{forward_reads_suffix}.trimmed.#{forward_reads_file_extension} -t #{trim_quality_cutoff} -l #{trim_point} -Q #{quality_scale} -v`
+    `#{fastq_quality_trimmer_path} -i #{directory}/#{sample_file_prefix}#{reverse_reads_suffix}.#{reverse_reads_file_extension} -o #{directory}/#{sample_file_prefix}#{reverse_reads_suffix}.trimmed.#{reverse_reads_file_extension} -t #{trim_quality_cutoff} -l #{trim_point} -Q #{quality_scale} -v`
+    `perl /tmp/fastq-remove-orphans.pl -1 #{sample_file_prefix}#{forward_reads_suffix}.trimmed.#{forward_reads_file_extension} -2 #{sample_file_prefix}#{reverse_reads_suffix}.trimmed.#{reverse_reads_file_extension}`
   end
 
   #  quake correction
   # write file for quake
   sample_map.each do |sample_file_prefix, sample_name|
+    puts "Error correcting files for #{sample_name}"
     output_file = File.open("quake_file_list.txt","w")
     output_file.puts "paired_#{sample_file_prefix}#{forward_reads_suffix}.trimmed.#{forward_reads_file_extension} paired_#{sample_file_prefix}#{reverse_reads_suffix}.trimmed.#{reverse_reads_file_extension}"
     output_file.close
     # run quake
-    system("#{quake_path} -f quake_file_list.txt -k 15 -q #{quality_scale}")
+    `#{quake_path} -f quake_file_list.txt -k 15 -q #{quality_scale}`
   end
   sample_map.each do |sample_file_prefix, sample_name|
-    system("perl /Volumes/NGS2_DataRAID/projects/MRSA/scripts/fastq-remove-orphans.pl -1 paired_#{sample_file_prefix}#{forward_reads_suffix}.trimmed.cor.#{forward_reads_file_extension} -2 paired_#{sample_file_prefix}#{reverse_reads_suffix}.trimmed.cor.#{reverse_reads_file_extension}")
+    `perl /Volumes/NGS2_DataRAID/projects/MRSA/scripts/fastq-remove-orphans.pl -1 paired_#{sample_file_prefix}#{forward_reads_suffix}.trimmed.cor.#{forward_reads_file_extension} -2 paired_#{sample_file_prefix}#{reverse_reads_suffix}.trimmed.cor.#{reverse_reads_file_extension}`
   end
 
   # cleanup and rename files
@@ -43,4 +47,18 @@ def trim_and_correct_fastqs(sample_map, directory, forward_reads_suffix, forward
     system("mv paired_paired_#{sample_file_prefix}#{forward_reads_suffix}.trimmed.cor.#{forward_reads_file_extension} #{sample_file_prefix}#{forward_reads_suffix}.trimmed.cor.#{forward_reads_file_extension}")
     system("mv paired_paired_#{sample_file_prefix}#{reverse_reads_suffix}.trimmed.cor.#{reverse_reads_file_extension} #{sample_file_prefix}#{reverse_reads_suffix}.trimmed.cor.#{reverse_reads_file_extension}")
   end
+end
+
+def calculate_read_length(filename)
+  read_length = nil
+  File.open(filename) do |f|
+    f.each do |line|
+     line.chomp!
+     if line =~ /^[GATCgatc]/
+       read_length = line.size
+       break
+     end
+    end
+  end
+  return read_length - 1
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: fastq-factory
 version: !ruby/object:Gem::Version
-  version: 0.1.3
+  version: 0.1.4
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-08-23 00:00:00.000000000 Z
+date: 2012-08-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: trollop
@@ -165,7 +165,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: -3475117921358810705
+      hash: -2858636493634267725
 required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
   requirements: