dna 0.0.4 → 0.0.5

data/lib/dna.rb

@@ -4,11 +4,12 @@
 class Dna # iterator
   include Enumerable
 
+  attr_reader :format
+
   def initialize(handle, args={})
     @handle = handle
-    @format = args[:format].to_sym
-
-    @iterator = 
+    @format = detect_format
+    @iterator =
       case @format
       when :fasta
         fasta_parser
@@ -21,6 +22,22 @@ class Dna # iterator
       end
   end
 
+  def detect_format
+    first_line = @handle.first
+    @handle.rewind if @handle.class == File
+
+    # detect qseq by counting number of tabs.
+    if first_line.split("\t").length == 11
+      return :qseq
+    elsif first_line[0].chr == '>'
+      return :fasta
+    elsif first_line[0].chr == '@'
+      return :fastq
+    else
+      raise Exception, "cannot detect format of input"
+    end
+  end
+
   def each &block
     @iterator.each do |r|
       if block_given?
@@ -38,21 +55,21 @@ class Dna # iterator
     Enumerator.new do |enum|
       @handle.each do |line|
         if line[0].chr == '>'
-          enum.yield Fasta.new(name: header, sequence: sequence) if sequence
+          enum.yield Fasta.new(:name => header, :sequence => sequence) if sequence
           sequence = ''
           header = line[1..-1].strip
         else
           sequence << line.strip.tr(' ','')
         end
       end
-      enum.yield Fasta.new(name: header, sequence: sequence)
+      enum.yield Fasta.new(:name => header, :sequence => sequence)
     end
   end
 
   def fastq_parser
     c = (0..3).cycle
     Enumerator.new do |enum|
-      params = { name: nil, sequence: nil, quality: nil }
+      params = { :name => nil, :sequence => nil, :quality => nil }
       @handle.each do |line|
         n = c.next
         case n
@@ -76,17 +93,17 @@ class Dna # iterator
       @handle.each do |line|
         line = line.strip.split("\t")
         record = QSEQ.new(
-          machine: line[0],
-          run: line[1],
-          lane: line[2],
-          tile: line[3],
-          x: line[4],
-          y: line[5],
-          index: line[6],
-          read_no: line[7],
-          sequence: line[8],
-          quality: line[9],
-          filtered: line[10],
+          :machine => line[0],
+          :run => line[1],
+          :lane => line[2],
+          :tile => line[3],
+          :x => line[4],
+          :y => line[5],
+          :index => line[6],
+          :read_no => line[7],
+          :sequence => line[8],
+          :quality => line[9],
+          :filtered => line[10]
         )
         enum.yield record
       end

data/spec/dna_spec.rb

@@ -3,9 +3,21 @@ require 'spec_helper'
 describe Dna do
   include_context "parser stuff"
 
-  let(:fasta) { Dna.new @fasta_file, format: 'fasta' }
-  let (:fastq) { Dna.new @fastq_file, format: 'fastq' }
-  let (:qseq) { Dna.new @qseq_file, format: 'qseq' }
+  let(:fasta) { Dna.new @fasta_file }
+  let (:fastq) { Dna.new @fastq_file }
+  let (:qseq) { Dna.new @qseq_file }
+
+  it 'can auto-detect fasta format' do
+    fasta.format == 'fasta'
+  end
+
+  it 'can auto-detect fastq format' do
+    fastq.format == 'fastq'
+  end
+
+  it 'can auto-detect qseq format' do
+    qseq.format == 'qseq'
+  end
 
   it 'can properly parse a crappily formatted fasta file' do
     records = fasta.to_a

data/spec/record_spec.rb

@@ -2,7 +2,7 @@ require 'spec_helper'
 
 describe Fasta do
 
-  let (:fasta) { Fasta.new name: 'test', sequence: 'GATC'}
+  let (:fasta) { Fasta.new :name => 'test', :sequence => 'GATC'}
 
   it 'can be created' do
     fasta.should_not be_nil
@@ -26,7 +26,7 @@ describe Fasta do
 end
 
 describe Fastq do
-  let (:fastq) { Fastq.new name: 'test', sequence: 'GATC', quality: 'BBBB' }
+  let (:fastq) { Fastq.new :name => 'test', :sequence => 'GATC', :quality  => 'BBBB' }
 
   it 'can be created' do
     fastq.should_not be_nil
@@ -58,17 +58,17 @@ describe QSEQ do
 
   let (:properties) {
     line = string.split("\t")
-    { machine: line[0],
-      run: line[1],
-      lane: line[2],
-      tile: line[3],
-      x: line[4],
-      y: line[5],
-      index: line[6],
-      read_no: line[7],
-      sequence: line[8],
-      quality: line[9],
-      filtered: line[10] }
+    { :machine => line[0],
+      :run => line[1],
+      :lane => line[2],
+      :tile => line[3],
+      :x => line[4],
+      :y => line[5],
+      :index => line[6],
+      :read_no => line[7],
+      :sequence => line[8],
+      :quality => line[9],
+      :filtered => line[10] }
   }
 
   let (:qseq) {

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dna
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.0.5
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2010-02-07 00:00:00.000000000 Z
+date: 2010-02-10 00:00:00.000000000 Z
 dependencies: []
 description: for parsing various types of DNA sequence files
 email: harekrishna@gmail.com