X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fassemble_seq_idba;h=87b81a2675769dfa0dc411a48bea9d4303df3383;hb=2f0fd91b461033529a4a72e161bd133252a22eb6;hp=861e0380ee6410ab731c51e211221ac776ebb2bf;hpb=d85ba667e3b8a39f11e47a8d0d3c00d80485441d;p=biopieces.git

diff --git a/bp_bin/assemble_seq_idba b/bp_bin/assemble_seq_idba
index 861e038..87b81a2 100755
--- 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
@@ -24,61 +24,76 @@
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-# Assemble sequences in the stream using IDBA.
+# Assemble sequences in the stream using IDBA-UD.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-require 'biopieces'
-require 'fasta'
-require 'pp'
-
-ok_methods = "ublast,usearch,uclust,usearch_uclust"
+require 'maasha/biopieces'
+require 'maasha/fasta'
 
 casts = []
-casts << {:long=>'name',        :short=>'n', :type=>'string', :mandatory=>true,  :default=>nil, :allowed=>nil, :disallowed=>nil}
-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=>'k_value_min', :short=>'k', :type=>'uint',   :mandatory=>false, :default=>25,  :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'k_value_max', :short=>'K', :type=>'uint',   :mandatory=>false, :default=>50,  :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=>'prefix_len',  :short=>'P', :type=>'uint',   :mandatory=>false, :default=>3,   :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'no_stream',   :short=>'x', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-
-bp = Biopieces.new
+casts << {:long=>'directory',  :short=>'d', :type=>'dir',  :mandatory=>true,  :default=>nil, :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=>'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=>'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 = bp.parse(ARGV, casts)
+options = Biopieces.options_parse(ARGV, casts)
 
 Dir.mkdir(options[:directory]) unless Dir.exists?(options[:directory])
 
-file_fasta = [options[:directory], options[:name] ].join(File::SEPARATOR) + ".fna"
-
-Fasta.open(file_fasta, mode="w") do |fasta_io|
-  bp.each_record do |record|
-    fasta_io.puts record
-  end
+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, "w") do |fasta_io|
+		input.each_record do |record|
+      if record[:SEQ_NAME] and record[:SEQ]
+        seq = Seq.new_bp(record)
+
+        fasta_io.puts seq.to_fasta
+      end
+		end
+	end
+
+	unless File.size(file_fasta) == 0
+		prefix = File.join(options[:directory], "IDBA-UD")
+
+		commands = []
+		commands << "nice -n 19"
+		commands << "idba_ud"
+		commands << "--read #{file_fasta}"
+		commands << "--out #{prefix}"
+		commands << "--mink #{options[:kmer_min]}"
+		commands << "--maxk #{options[:kmer_max]}"
+		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(" ")
+
+	  system(command)
+	  raise "Command failed: #{command}" unless $?.success?
+
+    file_contig = File.join(options[:directory], "IDBA-UD", "contig.fa")
+
+		Fasta.open(file_contig, "r") do |fasta_io|
+			fasta_io.each do |entry|
+				output.puts entry.to_bp
+			end
+		end
+	end
 end
 
-output = [options[:directory], options[:name] ].join(File::SEPARATOR)
-
-commands = []
-commands << "nice -n 19"
-commands << "idba"
-commands << "--read #{file_fasta}"
-commands << "--output #{output}"
-commands << "--scaffold" if options[:scaffold]
-commands << "--mink #{options[:k_value_min]}"
-commands << "--maxk #{options[:k_value_max]}"
-commands << "--minCount #{options[:count_min]}"
-commands << "--cover #{options[:cover]}"
-commands << "--minPairs #{options[:pairs_min]}"
-commands << "--prefixLength #{options[:prefix_len]}"
-commands << "> /dev/null 2>&1" unless options[:verbose]
-
-command = commands.join(" ")
-system(command)
-raise "Command failed: #{command}" unless $?.success?
+FileUtils.remove_entry_secure file_fasta
+FileUtils.remove_entry_secure options[:directory] if options[:clean]
+
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<