bio-samtools 2.0.5 → 2.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 67a8eba02e5b455c464caef319f561aef14aee9c
-  data.tar.gz: fb06f8ed5dc89daa39e4b30bd58d5c7bb87f1f27
+  metadata.gz: 0a798765b86732cd5e41c96a458439fbc51ec41c
+  data.tar.gz: 43315c48fc5fbc0f182e024031970f7e122b3446
 SHA512:
-  metadata.gz: 2a4b6c33c05059ccb6c938d1d730ce4b329d9af7d5c6b03d92756024d192d9e1a0c186a0ef491be95d163ee9b96bb1277f94855c02c25df7a0fdf646557064d2
-  data.tar.gz: de519c49fd975ee9a84fdca869d13c98c8759fd0e78a28ea0e9fd089ee7b10c90bdaf3f4eeedf9b5ddf0373b968197d54a1d81e5bc25c8f551ef54e897c6020c
+  metadata.gz: abe3c8c927c6459c67fc9a7c79437dad67a9e42c659cd04805de5cc5f8e3b2fc0725a7cb5906e36eb99331b8c3f7e2475d69b532689428b81b40af0361d99b58
+  data.tar.gz: eba5e3fff77e96192f8788eb0cf12d1c34cfd595d9954394d02c3780e45619aeb48065d5dbcea373e0febd6a8be2b179e020b2bbfc014cfa9e831005273908be

data/VERSION

@@ -1 +1 @@
-2.0.5
+2.1.0

data/bio-samtools.gemspec

@@ -2,17 +2,17 @@
 # DO NOT EDIT THIS FILE DIRECTLY
 # Instead, edit Jeweler::Tasks in Rakefile, and run 'rake gemspec'
 # -*- encoding: utf-8 -*-
-# stub: bio-samtools 2.0.5 ruby lib
+# stub: bio-samtools 2.1.0 ruby lib
 # stub: ext/mkrf_conf.rb
 
 Gem::Specification.new do |s|
   s.name = "bio-samtools"
-  s.version = "2.0.5"
+  s.version = "2.1.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.require_paths = ["lib"]
   s.authors = ["Ricardo Ramirez-Gonzalez", "Dan MacLean", "Raoul J.P. Bonnal"]
-  s.date = "2014-05-31"
+  s.date = "2014-09-03"
   s.description = "Binder of samtools for ruby, on the top of FFI. \n\n  This project was born from the need to add support of BAM files to \n  the gee_fu genome browser (http://github.com/danmaclean/gee_fu)."
   s.email = "ilpuccio.febo@gmail.com"
   s.extensions = ["ext/mkrf_conf.rb"]
@@ -86,7 +86,6 @@ Gem::Specification.new do |s|
     "ext/Rakefile",
     "ext/mkrf_conf.rb",
     "lib/bio-samtools.rb",
-    "lib/bio/.DS_Store",
     "lib/bio/BIOExtensions.rb",
     "lib/bio/db/alignment.rb",
     "lib/bio/db/fastadb.rb",
@@ -98,6 +97,7 @@ Gem::Specification.new do |s|
     "lib/bio/db/vcf.rb",
     "test/.gitignore",
     "test/helper.rb",
+    "test/sample.vcf",
     "test/samples/.gitignore",
     "test/samples/small/dupes.bam",
     "test/samples/small/dupes.sam",
@@ -132,13 +132,11 @@ Gem::Specification.new do |s|
     "test/samples/small/test_cov.svg",
     "test/samples/small/testu.bam",
     "test/samples/small/testu.bam.bai",
+    "test/samples/small/testu.bed",
     "test/test_bio-samtools.rb",
     "test/test_pileup.rb",
     "test/test_sam.rb",
     "test/test_vcf.rb",
-    "tutorial/images/out.svg",
-    "tutorial/images/out2.svg",
-    "tutorial/images/out3.svg",
     "tutorial/tutorial.html",
     "tutorial/tutorial.md",
     "tutorial/tutorial.pdf"

data/lib/bio/db/fastadb.rb

@@ -31,7 +31,7 @@ module Bio::DB::Fasta
     #The return object is of type Index. 
     def [](args)
       tmp = @entries[args]
-      new_index = Index.new
+      @new_index = Index.new
       tmp.each do | entry |
         @new_index << entry
       end       
@@ -69,21 +69,35 @@ module Bio::DB::Fasta
   class Region
     BASE_COUNT_ZERO =  {:A => 0, :C => 0, :G => 0,  :T => 0}
     attr_accessor :entry, :start, :end, :orientation
-    attr_accessor :pileup, :average_coverage, :reference, :base_ratios, :consensus, :coverages, :bases, :total_cov, :called
+
+    attr_accessor :pileup, :average_coverage, :snps, :reference, :allele_freq, :consensus, :coverages, :bases, :total_cov, :called
+  
+    def initialize(args ={})
+      @entry = args[:entry]
+      @start = args[:start]
+      @end = args[:end]
+      @orientation = args[:orientation]
+    end
+  
+
+  
 
     #TODO: Debug, as it hasnt been tested in the actual code. 
-    def base_ratios_for_base(base)
+    def allele_freq_for_base(base)
       @all_ratios = Hash.new unless @all_ratios
       unless @all_ratios[base]
         ratios = Array.new
         for i in (0..region.size-1)
-          ratios << @base_ratios[i][base]
+          ratios << @allele_freq[i][base]
         end
         @all_ratios[base] = ratios
       end
       @all_ratios[base]
     end
 
+    alias_method :base_ratios_for_base, :allele_freq_for_base
+    alias_method :base_ratios, :allele_freq
+
     #Calculates the concensus, base ratios, coverages and total coverages in the region
     #* min_cov minimum coverage to make a call (default 0)
     #* min_per minimum representation to make make a call. If more than one base 
@@ -94,7 +108,7 @@ module Bio::DB::Fasta
       self.called = 0
       reference = self.reference.downcase
 
-      self.base_ratios = Array.new(self.size, BASE_COUNT_ZERO) 
+      self.allele_freq = Array.new(self.size, BASE_COUNT_ZERO) 
       self.bases = Array.new(self.size, BASE_COUNT_ZERO) 
       self.coverages = Array.new(self.size, 0)
       self.total_cov = 0
@@ -102,7 +116,7 @@ module Bio::DB::Fasta
       self.pileup.each do | pile |
 
         if pile.coverage > min_cov
-          self.base_ratios[pile.pos - self.start ] = pile.base_ratios
+          self.allele_freq[pile.pos - self.start ] = pile.allele_freq
           reference[pile.pos - self.start   ] = pile.consensus_iuap(min_per).upcase
           self.coverages[pile.pos - self.start   ]  = pile.coverage.to_i
           self.bases[pile.pos - self.start       ]  = pile.bases
@@ -133,10 +147,7 @@ module Bio::DB::Fasta
       fields_2 = fields_1[1].split("-")
       raise FastaDBException.new(), "Invalid region. #{string}" if fields_1.length != 2 || fields_2.length != 2
 
-      reg = Region.new
-      reg.entry = fields_1[0]
-      reg.start = fields_2[0].to_i
-      reg.end = fields_2[1].to_i
+      reg = Region.new(:entry=> fields_1[0], :start=>fields_2[0].to_i, :end=>fields_2[1].to_i)
 
       if reg.end < reg.start 
         reg.orientation = :reverse
@@ -209,8 +220,6 @@ module Bio::DB::Fasta
 
     #The region needs to have a method to_region or a method to_s that ha the format "chromosome:start-end" as in samtools
     def fetch_sequence(region)
-
-
       query = region.to_s
       query = region.to_region.to_s if region.respond_to?(:to_region) 
       command = "#{@samtools} faidx #{@fasta_path} '#{query}'"

data/lib/bio/db/pileup.rb

@@ -85,14 +85,20 @@ module Bio
       def consensus
           if @consensus.nil?
             max = self.non_refs.values.max
+            #if the ref base is in more than half the coverage..
             if (self.ref_count / self.coverage) > 0.5
-              @consensus = self.ref_base 
-            elsif self.ref_count > max
+              #..then the ref base is the concensus
               @consensus = self.ref_base
+            ##not sure if the following will ever apply as the non_refs method also returns the ref base count, hence can never be over the max count  
+            #elsif self.ref_count > max
+            #  @consensus = self.ref_base
             else
+              #get the base(s) and count(s) that has the max count
               arr = self.non_refs.select {|k,v| v == max }
+              #just get the bases (remove the counts)
               bases = arr.collect {|b| b[0].to_s }
-              bases << self.ref_base if self.ref_count == max
+              #add the ref base if the ref base has a max count (commenting this out as it should already be in)
+              #bases << self.ref_base if self.ref_count == max
               @consensus = bases.sort.join
             end
           end
@@ -104,18 +110,16 @@ module Bio
         alt,g = self.genotype_list 
         alt = self.consensus.split(//).join(',') unless self.ref_base == '*'
         alt = '.' if alt == self.ref_base
+        alt = alt.split(',')
+        #if the reference base is in alt, remove it
+        alt.delete(self.ref_base.to_s)
+        alt = alt.join(',')
         [self.ref_name, self.pos, '.', self.ref_base, alt, self.snp_quality.to_i, "0", "DP=#{self.coverage.to_i}", "GT:GQ:DP", "#{g}:#{self.consensus_quality.to_i}:#{self.coverage.to_i}" ].join("\t")
       end
       
       private
       def Pileup.vcf_header
-        %{##fileformat=VCFv3.3
-          ##INFO=DP,1,Integer,"Total Depth"
-          ##FORMAT=GT,1,String,"Genotype"
-          ##FORMAT=GQ,1,Integer,"Genotype Quality"
-          ##FORMAT=DP,1,Integer,"Read Depth"
-          #CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tDATA
-        }.join("\n")
+        %{##fileformat=VCFv3.3\n##INFO=DP,1,Integer,"Total Depth"\n##FORMAT=GT,1,String,"Genotype"\n##FORMAT=GQ,1,Integer,"Genotype Quality"\n##FORMAT=DP,1,Integer,"Read Depth"\n#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tDATA\n}
       end
       
       
@@ -158,7 +162,7 @@ module Bio
         elsif bases[1] == self.ref_base
            return [bases[0],'0/1']
         else
-          return ["#{bases[0]},#{bases[1]}",'1/2']
+          return ["#{bases[0]},#{bases[1]}",'1/1']
         end 
       end
       
@@ -216,7 +220,7 @@ module Bio
          #puts self.ref_count
          @bases[self.ref_base.upcase.to_sym] = self.ref_count 
          @bases
-       end
+      end
 
        def base_coverage
          total = 0
@@ -226,34 +230,38 @@ module Bio
          total
        end
 
-       def base_ratios
-         return @base_ratios if @base_ratios
+      #returns the frequency of all bases in pileup position
+       def allele_freq
+         return @allele_frequency if @allele_frequency
          bases = self.bases
-         @base_ratios = Hash.new
+         @allele_frequency = Hash.new
          bases.each do |k,v| 
-           @base_ratios[k] = v.to_f/self.base_coverage.to_f 
+           @allele_frequency[k] = v.to_f/self.base_coverage.to_f 
          end
-         @base_ratios
+         @allele_frequency
        end
 
        # returns the consensus (most frequent) base from the pileup, if there are equally represented bases returns a string of all equally represented bases in alphabetical order   
        def consensus_iuap(minumum_ratio_for_iup_consensus)
          
+         tmp = []
          if @consensus_iuap.nil?
            @consensus_iuap = self.ref_base.downcase
            bases = self.bases
-           tmp = String.new
+           #tmp = String.new
            bases.each do |k,v|
              tmp << k[0].to_s if v/self.coverage.to_f > minumum_ratio_for_iup_consensus
            end
            if tmp.length > 0
-             @consensus_iuap = Bio::NucleicAcid.to_IUAPC(tmp)
+             tmp = tmp.collect{ |x| Bio::Sequence::NA.new(x) }
+              # creates alignment object
+              a = Bio::Alignment.new(tmp)
+              # shows IUPAC consensus
+             @consensus_iuap = a.consensus_iupac
            end
          end 
          @consensus_iuap
        end
-       
-       
     end
   end
 end

data/lib/bio/db/sam.rb

@@ -78,8 +78,9 @@ module Bio
           opts['1'] = opts[:one]
           opts.delete(:one)
         end
-
-        command = form_opt_string(@samtools, 'view', opts, [:b, :h, :H, :S, :u, '1', :x, :X, :c, :B]) + " " + region
+        command = String.new
+        command = form_opt_string(@samtools, 'view', opts, [:b, :h, :H, :S, :u, '1', :x, :X, :c, :B]) 
+        commad = command + " '#{region}'" if region.size > 0
         @last_command = command
         type = (opts[:u] or opts[:b]) ? :binary : :text
         klass = (type == :binary) ? String : Bio::DB::Alignment
@@ -125,6 +126,9 @@ module Bio
       #* bin - the amount of bins to split the histogram into. The arithmetic mean score for each bin will be plotted. [default 30 bins]
       #* svg - a file to write the svg image to [default a String object containing the SVG]
       def plot_coverage(chr,start,length, opts={})
+        chr = opts[:chr] if chr.nil?
+        start = opts[:start] if start.nil?
+        length = opts[:length] if length.nil?
         if opts[:bin]
           bin = length/opts[:bin]
         else
@@ -141,7 +145,7 @@ module Bio
         :font_size => 14
         )
         default_options = {:glyph => :histogram, 
-        :stroke_color => 'black',
+        :stroke => 'black',
         :fill_color => 'gold',
         :track_height => 150,
         :name => 'read coverage', 
@@ -264,7 +268,7 @@ module Bio
         end
 
         command = form_opt_string(@samtools, "mpileup", opts, [:R, :B, :E, "6", :A, :g, :u, :I] )
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         if opts[:u]
           command = command + " | #{@bcftools} view -cg -"
         end
@@ -287,7 +291,7 @@ module Bio
           seq = "n" * (stop-start) 
         else
           command = "#{@samtools} faidx #{@fasta} '#{chr}:#{start}-#{stop}'"
-          puts command if $VERBOSE
+          puts stderr.read if $VERBOSE
           @last_command = command
           seq = ""
           yield_from_pipe(command, String, :text ) {|line| seq = seq + line unless line =~ /^>/}
@@ -319,7 +323,7 @@ module Bio
       #* out_index - [STRING] name of index
       def index(opts={})
         command = "#{@samtools} index #{@bam} #{opts[:out_index]}"
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         system(command)
       end
@@ -334,7 +338,7 @@ module Bio
           remove_reads = "-r"
         end
         command = "#{@samtools} fixmate #{remove_reads} #{@bam} #{opts[:out_bam]}"
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         system(command)
       end
@@ -344,7 +348,7 @@ module Bio
       #generate simple stats with regard to the number and pairing of reads mapped to a reference
       def flag_stats(opts={})
         command = form_opt_string(@samtools, "flagstat", opts, [])
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         strings = []
         yield_from_pipe(command,String) {|line| strings << line.chomp}
@@ -359,6 +363,7 @@ module Bio
         stats = {}   
         command = form_opt_string(@samtools, "idxstats", {}, [])
         @last_command = command
+        puts stderr.read if $VERBOSE
         yield_from_pipe(command, String, :text, true, "#") do |line|
           info = line.chomp.split(/\t/)
           stats[ info[0] ] = {:length => info[1].to_i, :mapped_reads => info[2].to_i, :unmapped_reads => info[3].to_i }
@@ -383,7 +388,7 @@ module Bio
           reg.start = 1
           reg.end = v[:length]
           reg.orientation = :forward
-          @regions << reg unless @regions[k]
+          @regions[k] = reg unless @regions[k]
           yield reg if block_given?
         end
         @regions
@@ -439,7 +444,7 @@ module Bio
         command = "#{@samtools} merge #{options} #{out} #{bam_list}"
 
         @last_command = command
-        puts command puts command if $VERBOSE
+        puts command puts stderr.read if $VERBOSE
         system(command)
 
       end
@@ -449,9 +454,6 @@ module Bio
       #* out -[FILE] out file name
       #* bams -[FILES] or Bio::DB::Sam list of input bams, or Bio::DB::Sam objects
       def cat(opts={})
-        out = opts[:out]
-        opts.delete(:out)
-
         bam_list = opts[:bams].collect do |b|
           b.bam rescue b
         end.join(' ')
@@ -503,7 +505,7 @@ module Bio
         command = form_opt_string(@samtools, "sort", opts, [:n, :f, :o])
         command = command + " " + prefix
         @last_command = command
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         if opts[:o]
           yield_from_pipe(command, Bio::DB::Alignment)
         else
@@ -529,7 +531,7 @@ module Bio
           opts.delete(:s)
         end
         command = "#{form_opt_string(@samtools, "tview", opts)}"
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         system(command)
       end
@@ -544,7 +546,7 @@ module Bio
         else
           command = "#{@samtools} reheader #{header_sam} #{@bam}"
         end
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         system(command)
       end
@@ -560,7 +562,7 @@ module Bio
       #* E - Extended BAQ calculation. This option trades specificity for sensitivity, though the effect is minor. 
       def calmd(opts={}, &block)
         command = form_opt_string(@samtools, "calmd",  opts, [:E, :e, :u, :b, :S, :r] )+ " " + @fasta
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         type = :text 
         klass = Bio::DB::Alignment
@@ -581,7 +583,7 @@ module Bio
         end
 
         command = "#{form_opt_string(@samtools, "targetcut", opts, [] )}"
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         system(command)
       end
@@ -595,7 +597,7 @@ module Bio
       #* Q - [INT] Minimum base quality to be used in het calling. [13] 
       def phase(opts={})
         command = "#{form_opt_string(@samtools, "phase", opts, [:A, :F] )}"
-        puts command if $VERBOSE
+        puts stderr.read if $VERBOSE
         @last_command = command
         system(command)
       end
@@ -610,11 +612,7 @@ module Bio
       def depth(opts={})
         command = form_opt_string(@samtools, "depth", opts)
         @last_command = command
-        puts command if $VERBOSE
-        yield_from_pipe(command, String) do |line|
-          yield line.split(/\t/)
-        end
-
+        system(command)
       end
 
       #Returns the pipelup of a region, encapsulated as a Bio::DB::Fasta::Region object.  
@@ -668,6 +666,20 @@ module Bio
         end
       end
 
+      def bedcov(opts={})
+        bed = opts[:bed]
+        #bam = opts[:bam]
+        if opts.has_key?(:out)
+          out=opts[:out]
+          command = "#{@samtools} bedcov #{bed} #{@bam} > #{out}"
+        else
+          command = "#{@samtools} bedcov #{bed} #{@bam}"
+        end
+        #puts stderr.read if $VERBOSE
+        #puts command
+        @last_command = command
+        system(command)
+      end
 
 
       #Extract the reads that align to a region
@@ -740,7 +752,7 @@ module Bio
         "#{prog} #{command} #{opts_string} #{@bam}"
       end
 
-      # turns an opts hash into a s
+      # turns an opts hash into a string
       def commandify(opts, singles)
         list = []
         opts.each_pair do |tag,value|