upgraded uclust_seq to support usearch7

[biopieces.git] / code_ruby / lib / maasha / usearch.rb

diff --git a/code_ruby/lib/maasha/usearch.rb b/code_ruby/lib/maasha/usearch.rb
index 837585010f41fda3d18ea91292bdef452ba3076d..0bc72352d648c92d958a82aed390b3118518d4d2 100644 (file)
--- a/code_ruby/lib/maasha/usearch.rb
+++ b/code_ruby/lib/maasha/usearch.rb
@@ -27,8 +27,6 @@ require 'maasha/fasta'
 # Error class for all exceptions to do with Usearch.
 class UsearchError < StandardError; end
 
-SORT_LIMIT = 2_000_000_000   # use mergesort for files biggern than 2Gb.
-
 class Usearch
   include Enumerable
 
@@ -43,13 +41,9 @@ class Usearch
 
   # Method that calls Usearch sorting for sorting a FASTA file
   # according to decending sequence length.
-  def sort_length
-    # usearch -sort seqs.fasta -output seqs.sorted.fasta
-    if File.size(@infile) < SORT_LIMIT
-      @command << "usearch --sort #{@infile} --output #{@infile}.sort"
-    else
-      @command << "usearch --mergesort #{@infile} --output #{@infile}.sort"
-    end
+  def sortbylength
+    # usearch -sortbylength seqs.fasta -output seqs_sorted.fasta -minseqlength 64
+    @command << "usearch -sortbylength #{@infile} -output #{@infile}.sort"
 
                execute
 
@@ -58,20 +52,31 @@ class Usearch
 
   # Method that calls Usearch sorting for sorting a FASTA file
   # according to cluster size.
-  def sort_size
-    @command << "usearch --sortsize #{@infile} --output #{@infile}.sort"
+  def sortbysize
+    # usearch -sortbysize seqs.fasta -output seqs_sorted.fasta -minsize 4
+    @command << "usearch -sortbysize #{@infile} -output #{@infile}.sort"
 
                execute
 
     File.rename "#{@infile}.sort", @infile
   end
 
-       # Method to execute clustering de novo.
-  def cluster
-    @command << "usearch --cluster #{@infile} --uc #{@outfile} --id #{@options[:identity]}"
-    @command << "--rev" if @options[:comp]
+  # Method to execute cluster_fast.
+  def cluster_fast
+    # usearch -cluster_fast query.fasta -id 0.9 -centroids nr.fasta -uc clusters.uc
+    @command << "usearch -cluster_fast #{@infile} -id #{@options[:identity]} -uc #{@outfile}"
 
-               execute
+    execute
+  end
+
+  # Method to execute cluster_smallmem.
+  # NB sequences must be sorted with sortbylength or sortbysize.
+  def cluster_smallmem
+    # usearch -cluster_smallmem query.fasta -id 0.9 -centroids nr.fasta -uc clusters.uc
+    @command << "usearch -cluster_smallmem #{@infile} -id #{@options[:identity]} -uc #{@outfile}"
+    @command << "-strand both" if @options[:comp]
+
+    execute
   end
 
   # Method to execute database search.
@@ -188,6 +193,8 @@ class Usearch
 
       yield Align.new(entries) unless entries.empty?
     end
+
+    self # conventionally
   end
 
   private
@@ -209,7 +216,6 @@ class Usearch
        # 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 << "--quiet" unless @options[:verbose]
                command = @command.join(" ")
     $stderr.puts "Running command: #{command}" if @options[:verbose]