upgraded uclust_seq to support usearch7

[biopieces.git] / bp_bin / uclust_seq

diff --git a/bp_bin/uclust_seq b/bp_bin/uclust_seq
index 936d642909eb06c2da7075913c81c54bd44ab5bd..f155bdf53f06328bfe5d0c62232a667d3ee45d2c 100755 (executable)
--- a/bp_bin/uclust_seq
+++ b/bp_bin/uclust_seq
@@ -33,9 +33,9 @@ require 'maasha/fasta'
 require 'maasha/usearch'
 
 casts = []
-casts << {:long=>'no_sort',  :short=>'n', :type=>'flag',   :mandatory=>false, :default=>nil,      :allowed=>nil,        :disallowed=>nil}
-casts << {:long=>'comp',     :short=>'c', :type=>'flag',   :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_sort',  :short=>'n', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'comp',     :short=>'c', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'identity', :short=>'i', :type=>'float',  :mandatory=>true,  :default=>0.9, :allowed=>nil, :disallowed=>nil}
 
 options = Biopieces.options_parse(ARGV, casts)
 
@@ -45,7 +45,7 @@ file_fasta   = File.join(tmpdir, "in.fna")
 file_uclust  = File.join(tmpdir, "out.uc")
 
 Biopieces.open(options[:stream_in], file_records) 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|
       output.puts record
 
@@ -57,13 +57,14 @@ Biopieces.open(options[:stream_in], file_records) do |input, output|
 end
 
 us = Usearch.new(file_fasta, file_uclust, options)
-us.sort_length unless options[:no_sort]
-us.cluster
+
+us.sortbylength unless options[:no_sort]
+us.cluster_smallmem
 
 hash = {}
 
-us.each_cluster do |record|
-  hash[record[:Q_ID].to_sym] = record.dup
+us.each_cluster do |cluster|
+  hash[cluster[:Q_ID].to_sym] = cluster.dup
 end
 
 Biopieces.open(file_records, options[:stream_out]) do |input, output|