adding bzip2 support in ruby

[biopieces.git] / bp_bin / assemble_seq_idba

diff --git a/bp_bin/assemble_seq_idba b/bp_bin/assemble_seq_idba
index cdb6d9989c472a9713bb22bd34cdf8ec8491d4d9..87b81a2675769dfa0dc411a48bea9d4303df3383 100755 (executable)
--- a/bp_bin/assemble_seq_idba
+++ b/bp_bin/assemble_seq_idba
@@ -1,6 +1,6 @@
 #!/usr/bin/env ruby
 
-# Copyright (C) 2007-2011 Martin A. Hansen.
+# Copyright (C) 2007-2012 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -24,7 +24,7 @@
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-# Assemble sequences in the stream using IDBA.
+# Assemble sequences in the stream using IDBA-UD.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
@@ -33,73 +33,57 @@ require 'maasha/fasta'
 
 casts = []
 casts << {:long=>'directory',  :short=>'d', :type=>'dir',  :mandatory=>true,  :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'scaffold',   :short=>'s', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'kmer_min',   :short=>'k', :type=>'uint', :mandatory=>false, :default=>25,  :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'kmer_max',   :short=>'K', :type=>'uint', :mandatory=>false, :default=>50,  :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'kmer_min',   :short=>'k', :type=>'uint', :mandatory=>false, :default=>20,  :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'kmer_max',   :short=>'K', :type=>'uint', :mandatory=>false, :default=>100, :allowed=>nil, :disallowed=>nil}
 casts << {:long=>'count_min',  :short=>'c', :type=>'uint', :mandatory=>false, :default=>2,   :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'cover',      :short=>'C', :type=>'uint', :mandatory=>false, :default=>0,   :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'pairs_min',  :short=>'p', :type=>'uint', :mandatory=>false, :default=>5,   :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'pairs_min',  :short=>'p', :type=>'uint', :mandatory=>false, :default=>3,   :allowed=>nil, :disallowed=>nil}
 casts << {:long=>'prefix_len', :short=>'P', :type=>'uint', :mandatory=>false, :default=>3,   :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'clean',      :short=>'x', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'cpus',       :short=>'C', :type=>'uint', :mandatory=>false, :default=>0,   :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)
 
 Dir.mkdir(options[:directory]) unless Dir.exists?(options[:directory])
 
-file_fasta = File.join(options[:directory], "IDBA") + ".fna"
+file_fasta = File.join(options[:directory], "IDBA-UD") + ".fna"
 
 count = 0
 
 Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
-       Fasta.open(file_fasta, mode="w") do |fasta_io|
+       Fasta.open(file_fasta, "w") do |fasta_io|
                input.each_record do |record|
-      if options[:scaffold] # we need to fix the sequence name for mate-pair IDBA
-        if records[:SEQ_NAME] =~ /1$/
-          record[:SEQ_NAME] = "read#{count}/1"
-        else
-          record[:SEQ_NAME] = "read#{count}/2"
-
-          count += 1
-        end
-      end
+      if record[:SEQ_NAME] and record[:SEQ]
+        seq = Seq.new_bp(record)
 
-                       fasta_io.puts record
+        fasta_io.puts seq.to_fasta
+      end
                end
        end
 
        unless File.size(file_fasta) == 0
-               prefix = File.join(options[:directory], "IDBA")
+               prefix = File.join(options[:directory], "IDBA-UD")
 
                commands = []
                commands << "nice -n 19"
-               commands << "idba"
+               commands << "idba_ud"
                commands << "--read #{file_fasta}"
-               commands << "--output #{prefix}"
-               commands << "--scaffold" if options[:scaffold]
+               commands << "--out #{prefix}"
                commands << "--mink #{options[:kmer_min]}"
                commands << "--maxk #{options[:kmer_max]}"
-               commands << "--minCount #{options[:count_min]}"
-               commands << "--cover #{options[:cover]}"
-               commands << "--minPairs #{options[:pairs_min]}"
-               commands << "--prefixLength #{options[:prefix_len]}"
+               commands << "--min_count #{options[:count_min]}"
+               commands << "--min_pairs #{options[:pairs_min]}"
+               commands << "--prefix #{options[:prefix_len]}"
+               commands << "--num_threads #{options[:cpus]}"
                commands << "> /dev/null 2>&1" unless options[:verbose]
 
                command = commands.join(" ")
 
-               begin
-                       system(command)
-                       raise "Command failed: #{command}" unless $?.success?
-               rescue
-                       $stderr.puts "Command failed: #{command}"
-               end
+         system(command)
+         raise "Command failed: #{command}" unless $?.success?
 
-    if options[:scaffold]
-                 file_contig = File.join(options[:directory], "IDBA") + "-contig-mate.fa"
-    else
-                 file_contig = File.join(options[:directory], "IDBA") + "-contig.fa"
-    end
+    file_contig = File.join(options[:directory], "IDBA-UD", "contig.fa")
 
-               Fasta.open(file_contig, mode="r") do |fasta_io|
+               Fasta.open(file_contig, "r") do |fasta_io|
                        fasta_io.each do |entry|
                                output.puts entry.to_bp
                        end