added requirement of RubyInline

[biopieces.git] / bp_bin / uclust_seq

diff --git a/bp_bin/uclust_seq b/bp_bin/uclust_seq
index bd63df8e3935b64e99cbca600a335b6e2ccb9082..b20a29c672673ce690870e7f1e31d54df24e30b0 100755 (executable)
--- a/bp_bin/uclust_seq
+++ b/bp_bin/uclust_seq
@@ -58,6 +58,13 @@ class Uclust
     File.rename "#{@infile}.sort", @infile
   end
 
+       # Method to execute clustering de novo.
+  def cluster
+    @command << "usearch --cluster #{@infile} --uc #{@outfile} --id #{@options[:identity]}"
+
+               execute
+  end
+
   # Method to execute database search.
   def usearch
     # usearch --query query.fasta --db db.fasta --uc results.uc --id 0.90 [--evalue E]
@@ -67,13 +74,6 @@ class Uclust
                execute
   end
 
-       # Method to execute clustering de novo.
-  def cluster
-    @command << "usearch --cluster #{@infile} --uc #{@outfile} --id #{@options[:identity]}"
-
-               execute
-  end
-
        # Method to execute clustering to database plus de novo if not matched.
   def usearch_uclust
     # usearch --cluster seqs_sorted.fasta --db db.fasta --uc results.uc --id 0.90
@@ -129,7 +129,7 @@ class Uclust
        end
 end
 
-ok_methods = "usearch,uclust,usearch_uclust"
+ok_methods = "uclust,usearch,usearch_uclust"
 
 casts = []
 casts << {:long=>'no_sort',  :short=>'n', :type=>'flag',   :mandatory=>false, :default=>nil,      :allowed=>nil,        :disallowed=>nil}
@@ -141,6 +141,10 @@ casts << {:long=>'e_val',    :short=>'e', :type=>'float',  :mandatory=>false, :d
 
 options = Biopieces.options_parse(ARGV, casts)
 
+# At high identities, around 96% and above, compressed indexes are often more sensitive, faster 
+# and use less RAM. Compressed indexes are disabled by default, so I generally recommend that 
+# you specify the --slots and --w options when clustering at high identities.
+
 tmpdir  = Biopieces.mktmpdir
 infile  = File.join(tmpdir, "in.fna")
 outfile = File.join(tmpdir, "out.uc")
@@ -157,13 +161,12 @@ Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
   end
 
   uclust = Uclust.new(infile, outfile, options)
-  uclust.sort unless options[:no_sort]
+  uclust.sort unless options[:no_sort] or options[:method] == "usearch"
 
   case options[:method].to_s
-  when "usearch"        then uclust.usearch
   when "uclust"         then uclust.cluster
+  when "usearch"        then uclust.usearch
   when "usearch_uclust" then uclust.usearch_uclust
-  else raise "Unknown method: #{options[:method]}"
   end
 
   uclust.each do |record|