lederhosen 0.0.9 → 0.0.10

data/lib/lederhosen/tasks/filter.rb

@@ -0,0 +1,79 @@
+##
+# FILTER READS WITH LOW ABUNDANCE KMERS
+#
+
+module Lederhosen
+  class CLI
+
+    desc "filter fasta file",
+         "--input=joined.fasta --output=filtered.fasta --k=10 --cutoff=50"
+
+    method_option :input,    :type => :string,  :required => true
+    method_option :output,   :type => :string,  :required => true
+    method_option :k,        :type => :numeric, :required => true
+    method_option :cutoff,   :type => :numeric, :required => true
+
+    def filter
+      input  = options[:input]
+      output = options[:output]
+      k_len  = options[:k].to_i
+      cutoff = options[:cutoff]
+
+      counting_table = Hash.new { |h, k| h[k] = 0 }
+      total_reads = 0
+
+      ohai "counting kmers"
+      File.open(input) do |handle|
+        records = Dna.new handle
+        records.each do |r|
+          total_reads += 1
+          kmers = r.sequence.to_kmers(k_len)
+          kmers.each { |x| counting_table[x] += 1 }
+        end
+      end
+
+      sum_of_kmers = counting_table.values.inject(:+)
+
+      ohai "total reads = #{total_reads}"
+      ohai "sum of kmers = #{sum_of_kmers}"
+
+      kept = 0
+      total_reads = total_reads.to_f
+      
+      pbar = ProgressBar.new "saving", total_reads.to_i
+      output = File.open(output, 'w')
+      File.open(input) do |handle|
+        records = Dna.new handle
+
+        records.each do |r|
+          kmers = r.sequence.to_kmers(k_len)
+
+          # check if any of the kmers are rare
+          keep = true
+          coverage = 0
+          kmers.each do |kmer|
+            # if any of the kmers are rare, don't print the read
+            c = counting_table[kmer]
+            coverage += c
+            if c < cutoff
+              keep = false
+              break
+            end
+          end
+      
+          if keep
+            kept += 1
+            output.puts r
+          end
+          pbar.inc      
+        end
+      end
+
+      pbar.finish
+      
+      ohai "survivors = #{kept} (#{kept/total_reads.to_f})"
+      output.close
+    end
+  end
+
+end

data/lib/lederhosen.rb

@@ -5,3 +5,11 @@ require 'set'
 require 'progressbar'
 
 Dir.glob(File.join(File.dirname(__FILE__), 'lederhosen', '*.rb')).each { |f| require f }
+
+class String
+  def to_kmers(k)
+    return [] if k == 0
+    k -= 1
+    (0..(self.length-k-1)).collect { |i| self[i..i+k] }
+  end
+end

data/lib/version.rb

@@ -1,3 +1,3 @@
 module Lederhosen
-  VERSION = '0.0.9'
+  VERSION = '0.0.10'
 end

data/spec/misc_spec.rb

@@ -0,0 +1,11 @@
+require 'spec_helper'
+
+describe String do
+  it 'generate_kmers should generate kmers for a string' do
+    'test'.to_kmers(2).should == ['te', 'es', 'st']
+    'test'.to_kmers(3).should == ['tes', 'est']
+    'test'.to_kmers(4).should == ['test']
+    'test'.to_kmers(5).should == []
+    'test'.to_kmers(0).should == []
+  end
+end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: lederhosen
 version: !ruby/object:Gem::Version 
-  hash: 13
+  hash: 11
   prerelease: 
   segments: 
   - 0
   - 0
-  - 9
-  version: 0.0.9
+  - 10
+  version: 0.0.10
 platform: ruby
 authors: 
 - Austin G. Davis-Richardson
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-05-01 00:00:00 Z
+date: 2012-05-14 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: dna
@@ -121,6 +121,7 @@ files:
 - lib/lederhosen/cli.rb
 - lib/lederhosen/helpers.rb
 - lib/lederhosen/tasks/cluster.rb
+- lib/lederhosen/tasks/filter.rb
 - lib/lederhosen/tasks/join.rb
 - lib/lederhosen/tasks/name.rb
 - lib/lederhosen/tasks/otu_table.rb
@@ -135,6 +136,7 @@ files:
 - spec/data/ILT_L_9_B_002_1.txt
 - spec/data/ILT_L_9_B_002_3.txt
 - spec/helpers_spec.rb
+- spec/misc_spec.rb
 - spec/pipeline_spec.rb
 - spec/spec_helper.rb
 homepage: http://github.com/audy/lederhosen
@@ -176,5 +178,6 @@ test_files:
 - spec/data/ILT_L_9_B_002_1.txt
 - spec/data/ILT_L_9_B_002_3.txt
 - spec/helpers_spec.rb
+- spec/misc_spec.rb
 - spec/pipeline_spec.rb
 - spec/spec_helper.rb