lederhosen 0.2.2 → 0.2.3

data/lib/lederhosen/tasks/uniquify.rb

@@ -0,0 +1,49 @@
+##
+# uniquify - uniquify a fasta file, also output table with md5 -> number of reads
+#
+
+require 'digest/md5'
+
+module Lederhosen
+  class CLI
+    desc 'uniquify',
+      'uniquify a fasta file and generate a table with md5 -> abundance'
+
+    method_option :input, :type     => :string, :required => true
+    method_option :output, :type    => :string, :required => true
+    method_option :table_out, :type => :string, :required => true
+
+    def uniquify
+      input     = options[:input]
+      output    = options[:output]
+      table_out = options[:table_out]
+
+      sequence_counts = Hash.new { |h, k| h[k] = 0 }
+
+      out = File.open(output, 'w')
+
+      File.open(input) do |handle|
+        Dna.new(handle).each do |record|
+          unless sequence_counts.has_key? record.sequence
+            out.puts record
+          end
+          sequence_counts[record.sequence] += 1
+        end
+      end
+
+      out.close
+
+      # write table
+      File.open(table_out, 'w') do |out|
+        sequence_counts.each_pair do |sequence, count|
+          digest = Digest::MD5.hexdigest(sequence)
+          out.puts "#{digest},#{count}"
+        end
+      end
+
+      kept = sequence_counts.keys.size
+      total = sequence_counts.values.inject(:+)
+      ohai "kept #{kept} out of #{total} reads (#{100*kept/total.to_f})"
+    end
+  end
+end

data/lib/version.rb

@@ -1,3 +1,3 @@
 module Lederhosen
-  VERSION = '0.2.2'
+  VERSION = '0.2.3'
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: lederhosen
 version: !ruby/object:Gem::Version 
-  hash: 19
+  hash: 17
   prerelease: 
   segments: 
   - 0
   - 2
-  - 2
-  version: 0.2.2
+  - 3
+  version: 0.2.3
 platform: ruby
 authors: 
 - Austin G. Davis-Richardson
@@ -95,6 +95,7 @@ files:
 - lib/lederhosen/tasks/trim.rb
 - lib/lederhosen/tasks/uc_filter.rb
 - lib/lederhosen/tasks/uc_stats.rb
+- lib/lederhosen/tasks/uniquify.rb
 - lib/version.rb
 - readme.md
 - spec/data/ILT_L_9_B_001_1.txt.gz