polishing on uclust_seq

git-svn-id: http://biopieces.googlecode.com/svn/trunk@1101 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_bin/uclust_seq b/bp_bin/uclust_seq
index 4738212426b60e9bc27cc9fbb4bd054299fd984d..36e40ad19951077eec267a1436c1457cc6d112d8 100755 (executable)
--- a/bp_bin/uclust_seq
+++ b/bp_bin/uclust_seq
@@ -24,7 +24,7 @@
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-# 
+# Run Uclust on sequences in the stream.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
@@ -35,62 +35,75 @@ require 'fasta'
 class Uclust
   include Enumerable
 
-  def initialize(infile, outfile)
+  def initialize(infile, outfile, options)
     @infile  = infile
     @outfile = outfile
+               @options = options
+               @command = []
   end
 
   # Method that calls Uclusts sorting for sorting a FASTA file
   # according to decending sequence length.
-  def sort(options)
-    command = "nice -n 19"
-    command << " uclust --sort #{@infile} --output #{@infile}.sort"
-    command << " > /dev/null 2>&1" unless options[:verbose]
-    system(command)
-    raise "Failed sorting file: #{@infile}" unless $?.success?
+  def sort
+    @command << "uclust --sort #{@infile} --output #{@infile}.sort"
+
+               execute
 
     File.rename "#{@infile}.sort", @infile
   end
 
-  def ublast(options)
+  def ublast
     # uclust --ublast query_seqs.fasta --db database.fasta --blast6out filename --evalue E
-    options[:e_val] = 10 if options[:e_val].is_nil?
-    command = "nice -n 19"
-    command << " uclust --ublast #{@infile} --db #{options[:database]} --uc #{@outfile} --evalue #{options[:e_val]}"
-    command << " > /dev/null 2>&1" unless options[:verbose]
-    system(command)
-    raise "Command failed: #{command}" unless $?.success?
+    @options[:e_val] = 10 if @options[:e_val].is_nil?
+    @command << "uclust --ublast #{@infile} --db #{@options[:database]} --uc #{@outfile} --evalue #{@options[:e_val]}"
+               @command << "--gapopen *I" if @options.has_key? :no_gaps
+
+               execute
   end
 
-  def usearch(options)
+  # Method to execute database search.
+  def usearch
     # uclust --query query.fasta --db db.fasta --uc results.uc --id 0.90 [--evalue E]
-    command = "nice -n 19"
-    command << " uclust --query #{@infile} --db #{options[:database]} --uc #{@outfile} --id #{options[:identity]}"
-    command << " --evalue #{options[:e_val]}" if options.has_key? :e_val
-    command << " > /dev/null 2>&1" unless options[:verbose]
-    system(command)
-    raise "Command failed: #{command}" unless $?.success?
+    @command << "uclust --query #{@infile} --db #{@options[:database]} --uc #{@outfile} --id #{@options[:identity]}"
+    @command << "--evalue #{@options[:e_val]}" if @options.has_key? :e_val
+               @command << "--gapopen *I" if @options.has_key? :no_gaps
+
+               execute
   end
 
-  def uclust(options)
+       # Method to execute clustering de novo.
+  def uclust
     # uclust --input seqs_sorted.fasta --uc results.uc --id 0.90
-    command = "nice -n 19"
-    command << " uclust --input #{@infile} --uc #{@outfile} --id #{options[:identity]}"
-    command << " > /dev/null 2>&1" unless options[:verbose]
-    system(command)
-    raise "Command failed: #{command}" unless $?.success?
+    @command << "uclust --input #{@infile} --uc #{@outfile} --id #{@options[:identity]}"
+               @command << "--gapopen *I" if @options.has_key? :no_gaps
+
+               execute
   end
 
-  def usearch_uclust(options)
+       # Method to execute clustering to database plus de novo if not matched.
+  def usearch_uclust
     # uclust --input seqs_sorted.fasta --lib db.fasta --uc results.uc --id 0.90
-    command = "nice -n 19"
-    command << " uclust --input #{@infile} --lib #{options[:database]} --uc #{@outfile} --id #{options[:identity]}"
-    command << " --lib #{options[:database]}" if options.has_key? :database
-    command << " > /dev/null 2>&1" unless options[:verbose]
+    @command << "uclust --input #{@infile} --lib #{@options[:database]} --uc #{@outfile} --id #{@options[:identity]}"
+    @command << "--lib #{@options[:database]}" if @options.has_key? :database
+               @command << "--gapopen *I" if @options.has_key? :no_gaps
+
+               execute
+  end
+
+       # Method to execute a command using a system() call.
+       # The command is composed of bits from the @command variable.
+       def execute
+               @command.unshift "nice -n 19"
+               @command << "> /dev/null 2>&1" unless @options[:verbose]
+               command = @command.join(" ")
     system(command)
     raise "Command failed: #{command}" unless $?.success?
-  end
 
+               @command = []
+       end
+
+       # Method to parse a Uclust .uc file and for each line of data
+       # yield a Biopiece record.
   def each
     record = {}
 
@@ -127,6 +140,7 @@ casts << {:long=>'no_sort',  :short=>'n', :type=>'flag',   :mandatory=>false, :d
 casts << {:long=>'method',   :short=>'m', :type=>'string', :mandatory=>true,  :default=>"uclust", :allowed=>ok_methods, :disallowed=>nil}
 casts << {:long=>'database', :short=>'d', :type=>'file!',  :mandatory=>false, :default=>nil,      :allowed=>nil,        :disallowed=>nil}
 casts << {:long=>'identity', :short=>'i', :type=>'float',  :mandatory=>true,  :default=>0.9,      :allowed=>nil,        :disallowed=>nil}
+casts << {:long=>'no_gaps',  :short=>'g', :type=>'flag',   :mandatory=>false, :default=>nil,      :allowed=>nil,        :disallowed=>nil}
 casts << {:long=>'e_val',    :short=>'e', :type=>'float',  :mandatory=>false, :default=>nil,      :allowed=>nil,        :disallowed=>nil}
 
 bp = Biopieces.new
@@ -144,14 +158,14 @@ Fasta.open(infile, mode="w") do |fasta_io|
   end
 end
 
-uclust = Uclust.new(infile, outfile)
-uclust.sort(options) unless options[:no_sort]
+uclust = Uclust.new(infile, outfile, options)
+uclust.sort unless options[:no_sort]
 
 case options[:method].to_s
-when "ublast"         then uclust.ublast(options)
-when "usearch"        then uclust.usearch(options)
-when "uclust"         then uclust.uclust(options)
-when "usearch_uclust" then uclust.usearch_uclust(options)  
+when "ublast"         then uclust.ublast
+when "usearch"        then uclust.usearch
+when "uclust"         then uclust.uclust
+when "usearch_uclust" then uclust.usearch_uclust
 else raise "Unknown method: #{options[:method]}"
 end