rewrote assemble_seq_velvet

git-svn-id: http://biopieces.googlecode.com/svn/trunk@2066 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_bin/assemble_seq_velvet b/bp_bin/assemble_seq_velvet
index d613967586368f3ff377be48987095cbb60ee6d9..3b023b213caca2664ce071c075af00a396ba7f6e 100755 (executable)
--- a/bp_bin/assemble_seq_velvet
+++ b/bp_bin/assemble_seq_velvet
@@ -38,67 +38,33 @@ class Velvet
     @verbose       = verbose
   end
 
-  def run_velveth(kmer_min, kmer_max, type)
-    @kmer_min = kmer_min
-    @kmer_max = kmer_max
-
-    kmer = @kmer_min
-
-    while kmer <= @kmer_max
-      dir_velveth = File.join(@directory, "Kmer_#{kmer}")
-
-      Dir.mkdir(dir_velveth)
-
-      commands = []
-      commands << "velveth"
-      commands << dir_velveth
-      commands << kmer
-      commands << "-#{type}"
-      commands << "-fasta"
-      commands << @sequence_file
-
-      execute(commands)
-
-      kmer += 2
-    end
+  def run_velveth(kmer_min, kmer_max, kmer_step, type)
+    commands = []
+    commands << "velveth"
+    commands << @directory
+    commands << "#{kmer_min},#{kmer_max},#{kmer_step}"
+    commands << "-#{type}"
+    commands << "-fasta"
+    commands << @sequence_file
+
+    execute(commands)
   end
 
   def run_velvetg(cov_cutoff, exp_cov, paired)
-    Dir.glob("#{@directory}/Kmer_*").each do |dir_velveth|
-      files_velveth = Dir.glob("#{dir_velveth}/*")
-
-      dir_velvetg = File.join(dir_velveth, "Cov_cutoff_#{cov_cutoff}")
-
-      Dir.mkdir(dir_velvetg)
-      FileUtils.cp_r files_velveth, dir_velvetg
-      
-      commands = []
-      commands << "velvetg"
-      commands << dir_velvetg
-      commands << "-cov_cutoff #{cov_cutoff}"
-      commands << "-exp_cov #{exp_cov}"
-      commands << "-ins_length #{ins_length}" if paired
-      commands << "-clean yes"
-
-      execute(commands)
-    end
-  end
-
-  def pick_best_assembly
-    list = []
-
-    Dir.glob("#{@directory}/Kmer_*/Cov_cutoff_*/contigs.fa").each do |file|
-      n50 = fasta_n50(file)
-      list << [file, n50]
-    end
-
-    list.sort_by { |e| e.last }.last.first
+    commands = []
+    commands << "velvetg"
+    commands << @directory
+    commands << "-cov_cutoff #{cov_cutoff}"
+    commands << "-exp_cov #{exp_cov}"
+    commands << "-ins_length #{ins_length}" if paired
+    commands << "-clean yes"
+
+    execute(commands)
   end
 
   private
 
   def execute(commands)
-    commands.unshift "nice -n 19"
     commands.push "> /dev/null 2>&1" unless @verbose
 
     command = commands.join(" ")
@@ -110,49 +76,26 @@ class Velvet
       $stderr.puts "Command failed: #{command}"
     end
   end
-
-  def fasta_n50(file)
-    total   = 0
-    lengths = []
-    count   = 0
-    n50     = 0
-
-    Fasta.open(file, mode="r") do |fasta_io|
-      fasta_io.each do |entry|
-        total   += entry.length
-        lengths << entry.length
-      end
-    end
-
-    lengths.sort.reverse.each do |length|
-      count += length
-
-      if count >= total * 0.50
-        n50 = length
-        break
-      end
-    end
-
-    n50
-  end
 end
 
-types       = 'short,shortPaired,long,longPaired'
+types = 'short,shortPaired,long,longPaired'
 
 casts = []
 casts << {:long=>'directory',  :short=>'d', :type=>'dir',    :mandatory=>true,  :default=>nil,     :allowed=>nil,   :disallowed=>nil}
 casts << {:long=>'type',       :short=>'t', :type=>'string', :mandatory=>true,  :default=>'short', :allowed=>types, :disallowed=>nil}
 casts << {:long=>'kmer_min',   :short=>'k', :type=>'uint',   :mandatory=>true,  :default=>19,      :allowed=>nil,   :disallowed=>nil}
 casts << {:long=>'kmer_max',   :short=>'K', :type=>'uint',   :mandatory=>true,  :default=>31,      :allowed=>nil,   :disallowed=>nil}
+casts << {:long=>'kmer_step',  :short=>'s', :type=>'uint',   :mandatory=>true,  :default=>2,       :allowed=>nil,   :disallowed=>nil}
 casts << {:long=>'cov_cutoff', :short=>'c', :type=>'float',  :mandatory=>true,  :default=>'auto',  :allowed=>nil,   :disallowed=>nil}
 casts << {:long=>'exp_cov',    :short=>'e', :type=>'float',  :mandatory=>true,  :default=>'auto',  :allowed=>nil,   :disallowed=>nil}
-casts << {:long=>'ins_length', :short=>'i', :type=>'uint',   :mandatory=>false, :default=>700,     :allowed=>nil,   :disallowed=>nil}
+casts << {:long=>'ins_length', :short=>'i', :type=>'uint',   :mandatory=>false, :default=>nil,     :allowed=>nil,   :disallowed=>nil}
 casts << {:long=>'clean',      :short=>'X', :type=>'flag',   :mandatory=>false, :default=>nil,     :allowed=>nil,   :disallowed=>nil}
 
 options = Biopieces.options_parse(ARGV, casts)
 
 raise ArgumentError, "kmer_min #{options[:kmer_min]} must be uneven." if options[:kmer_min].even?
 raise ArgumentError, "kmer_max #{options[:kmer_max]} must be uneven." if options[:kmer_max].even?
+raise ArgumentError, "kmer_step #{options[:kmer_step]} must be even." unless options[:kmer_step].even?
 raise ArgumentError, "kmer_min >= kmer_max: #{options[:kmer_min]} >= #{options[:kmer_max]}" unless options[:kmer_max] > options[:kmer_min]
 
 Dir.mkdir(options[:directory]) unless Dir.exists?(options[:directory])
@@ -171,9 +114,9 @@ Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
 
        unless File.size(file_fasta) == 0
                velvet = Velvet.new(options[:directory], file_fasta, options[:verbose])
-               velvet.run_velveth(options[:kmer_min], options[:kmer_max], options[:type])
+               velvet.run_velveth(options[:kmer_min], options[:kmer_max], options[:kmer_step], options[:type])
                velvet.run_velvetg(options[:cov_cutoff], options[:exp_cov], options[:type].match("Paired"))
-               file_contigs = velvet.pick_best_assembly
+               file_contigs = File.join(options[:directory], "contigs.fa")
 
                Fasta.open(file_contigs, mode="r") do |fasta_io|
                        fasta_io.each do |entry|