transrate 0.0.8 → 0.0.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ad3086d63ee5deaf6ed931c5df095e776e601e2e
-  data.tar.gz: 188ff6eb56896b42de4ec8e4cee31e23d8387bed
+  metadata.gz: e4a7687e1bc2071fe2f043245e1eff90742f4f1e
+  data.tar.gz: 61ef6386c15fe8d56485c8c5737d3283aefaede3
 SHA512:
-  metadata.gz: 949dd128cb648f481bd7f473e037d9873325407393981674888b04e1b07f9729a1da5cd9d1efb42fb55838cfabcd6c03e33b42b9cfd25f80168fe78175aac035
-  data.tar.gz: 985cdfc7335410d3c7bd7f1c086b8d0aa87bd66b3af6e8859cc86d000583cca4b8dd728fce76add4e37b14636cd30a04d1876181a3879ec473a11f58368aaa69
+  metadata.gz: 42ab6bc0454bd683798c5e9a1d93a7687fd3282bac182275901b89435dc3c90203dc3ffab12ad128fdf4ebab99702df816f3534917a061da6efdbd120e85cf9b
+  data.tar.gz: fed477db5ad8a33560bdf25a9bff9185638e49e418ccd5fcf9808235034abbc03ea1c7dded07520f4e4d0bff3f784146417a7e11c58cf938a4007bc71e7c15fc

data/.gitignore

@@ -9,6 +9,7 @@ lib/bundler/man
 pkg
 rdoc
 spec/reports
+test
 test/tmp
 test/version_tmp
 tmp

data/bin/transrate

@@ -39,7 +39,7 @@ EOS
 end
 
 def pretty_print_hash hash, width
-  hash.map{ |k, v| "#{k.to_s}#{" " * (width - (k.length + v.to_i.to_s.length))}#{v.to_i}" }.join("\n")
+  hash.map{ |k, v| "#{k.to_s}#{" " * (width - (k.length + v.to_f.round(2).to_s.length))}#{v.to_f.round(2)}" }.join("\n")
 end
 
 include Transrate
@@ -59,7 +59,8 @@ if (opts.left && opts.right)
   puts "\ncalculating read diagnostics..."
   t0 = Time.now
   read_metrics = ReadMetrics.new a
-  read_results = read_metrics.run(opts.left, opts.right)
+  read_metrics.run(opts.left, opts.right)
+  read_results = read_metrics.read_stats
   puts "...done in #{Time.now - t0} seconds"
 else
   puts "\nno reads provided, skipping read diagnostics"
@@ -68,7 +69,8 @@ end
 puts "\ncalculating comparative metrics..."
 t0 = Time.now
 comparative_metrics = ComparativeMetrics.new(a, r)
-comparative_results = comparative_metrics.run
+comparative_metrics.run
+comparative_results = comparative_metrics.comp_stats
 puts "...done in #{Time.now - t0} seconds"
 
 report_width = 30
@@ -93,3 +95,9 @@ if comparative_results
   puts "-" *  report_width
   puts pretty_print_hash(comparative_results, report_width)
 end
+
+transrater = Transrater.new(a, r, opts.left, opts.right)
+transrater.run(opts.left, opts.right)
+puts "\n\n"
+puts "Overall score #{transrater.assembly_score.to_f.round(2)}"
+puts "\n" + "-" * report_width

data/lib/transrate.rb

@@ -6,4 +6,7 @@ require 'transrate/read_metrics'
 require 'transrate/usearch'
 require 'transrate/rb_hit'
 require 'transrate/reciprocal_annotation'
-require 'transrate/comparative_metrics'
+require 'transrate/comparative_metrics'
+require 'transrate/metric'
+require 'transrate/dimension_reduce'
+require 'transrate/express'

data/lib/transrate/#assembly.rb#

@@ -0,0 +1,130 @@
+require 'bio'
+require 'bettersam'
+require 'csv'
+require 'forwardable'
+
+module Transrate
+
+  class Assembly
+
+    include Enumerable
+    extend Forwardable
+    def_delegators :@assembly, :each, :<<
+
+    attr_accessor :ublast_db
+    attr_accessor :orfs_ublast_db
+    attr_accessor :protein
+    attr_reader :assembly
+
+    # number of bases in the assembly
+    attr_writer :n_bases
+
+    # assembly filename
+    attr_accessor :file
+
+    # assembly n50
+    attr_reader :n50
+
+    # Reuturn a new Assembly.
+    #
+    # - +:file+ - path to the assembly FASTA file
+    def initialize file
+      @file = file
+      @assembly = []
+      @n_bases = 0
+      Bio::FastaFormat.open(file).each do |entry|
+        @n_bases += entry.length
+        @assembly << entry
+      end
+      @assembly.sort_by! { |x| x.length }
+    end
+
+    # Return a new Assembly object by loading sequences
+    # from the FASTA-format +:file+
+    def self.stats_from_fasta file
+      a = Assembly.new file
+      a.basic_stats
+    end
+
+    def run
+      stats = self.basic_stats
+      stats.each_pair do |key, value|
+        ivar = "@#{key.gsub(/ /, '_')}".to_sym
+        self.instance_variable_set(ivar, value)
+      end
+    end
+
+    # Return a hash of statistics about this assembly
+    def basic_stats
+      cumulative_length = 0.0
+      # we'll calculate Nx for all these x
+      x = [90, 70, 50, 30, 10]
+      x2 = x.clone
+      cutoff = x2.pop / 100.0
+      res = []
+      n1k = 0
+      n10k = 0
+      orf_length_sum = 0
+      @assembly.each do |s|
+        n1k += 1 if s.length > 1_000
+        n10k += 1 if s.length > 10_000
+        orf_length_sum += orf_length(s.seq)
+
+        cumulative_length += s.length
+        if cumulative_length >= @n_bases * cutoff
+          res << s.length
+          if x2.empty?
+            cutoff=1
+          else
+            cutoff = x2.pop / 100.0
+          end 
+        end
+      end
+
+      mean = cumulative_length / @assembly.size
+      ns = Hash[x.map { |n| "N#{n}" }.zip(res)]
+      {
+        "n_seqs" => @assembly.size,
+        "smallest" => @assembly.first.length,
+        "largest" => @assembly.last.length,
+        "n_bases" => @n_bases,
+        "mean_len" => mean,
+        "n_1k" => n1k,
+        "n_10k" => n10k,
+        "orf percent" => 300*orf_length_sum/(@assembly.size*mean)
+      }.merge ns
+    end
+
+    # finds longest orf in a sequence
+    def orf_length sequence
+      longest=0
+      (1..6).each do |frame|
+        translated = Bio::Sequence::NA.new(sequence).translate(frame)
+        translated.split(/\*/).each do |orf|
+          if orf.length > longest
+            longest=orf.length
+          end
+        end
+      end
+      return longest
+    end
+
+    # return the number of bases in the assembly, calculating
+    # from the assembly if it hasn't already been done.
+    def n_bases
+      unless @n_bases
+        @n_bases = 0
+        @assembly.each { |s| @n_bases += s.length }
+      end
+      @n_bases
+    end
+
+    def print_stats
+      self.basic_stats.map do |k, v| 
+        "#{k}#{" " * (20 - (k.length + v.to_i.to_s.length))}#{v.to_i}"
+      end.join("\n")
+    end
+
+  end # Assembly
+
+end # Transrate

data/lib/transrate/assembly.rb

@@ -14,6 +14,7 @@ module Transrate
     attr_accessor :ublast_db
     attr_accessor :orfs_ublast_db
     attr_accessor :protein
+    attr_reader :assembly
 
     # number of bases in the assembly
     attr_writer :n_bases
@@ -63,18 +64,23 @@ module Transrate
       res = []
       n1k = 0
       n10k = 0
+      orf_length_sum = 0
       @assembly.each do |s|
-        new_cum_len = cumulative_length + s.length
-        prop = new_cum_len / self.n_bases
         n1k += 1 if s.length > 1_000
         n10k += 1 if s.length > 10_000
-        if prop >= cutoff
+        orf_length_sum += orf_length(s.seq)
+
+        cumulative_length += s.length
+        if cumulative_length >= @n_bases * cutoff
           res << s.length
-          break if x2.empty?
-          cutoff = x2.pop / 100.0
+          if x2.empty?
+            cutoff=1
+          else
+            cutoff = x2.pop / 100.0
+          end 
         end
-        cumulative_length = new_cum_len
       end
+
       mean = cumulative_length / @assembly.size
       ns = Hash[x.map { |n| "N#{n}" }.zip(res)]
       {
@@ -84,10 +90,25 @@ module Transrate
         "n_bases" => @n_bases,
         "mean_len" => mean,
         "n_1k" => n1k,
-        "n_10k" => n10k
+        "n_10k" => n10k,
+        "orf percent" => 300*orf_length_sum/(@assembly.size*mean)
       }.merge ns
     end
 
+    # finds longest orf in a sequence
+    def orf_length sequence
+      longest=0
+      (1..6).each do |frame|
+        translated = Bio::Sequence::NA.new(sequence).translate(frame)
+        translated.split(/\*/).each do |orf|
+          if orf.length > longest
+            longest=orf.length
+          end
+        end
+      end
+      return longest
+    end
+
     # return the number of bases in the assembly, calculating
     # from the assembly if it hasn't already been done.
     def n_bases

data/lib/transrate/bowtie2.rb

@@ -17,12 +17,12 @@ module Transrate
     def map_reads file, left, right=nil, insertsize=200, insertsd=50, outputname=nil
       lbase = File.basename(left)
       rbase = File.basename(right)
-      outputname ||= "#{lbase}.#{rbase}.sam"
+      outputname ||= "#{lbase}.#{rbase}.#{File.basename(file)}.sam"
       realistic_dist = insertsize + (3 * insertsd)
       unless File.exists? outputname
         # construct bowtie command
-        bowtiecmd = "#{@bowtie2} -k 3 -p 8 -X #{realistic_dist}"
-        bowtiecmd += " --no-unal --local --quiet"
+        bowtiecmd = "#{@bowtie2} --very-sensitive-local -p 8 -X #{realistic_dist}" # TODO number of cores should be variable '-p 8'
+        bowtiecmd += " --no-unal"
         bowtiecmd += " #{File.basename(file)} -1 #{left}"
         # paired end?
         bowtiecmd += " -2 #{right}" if right
@@ -41,4 +41,4 @@ module Transrate
 
   end # Bowtie2
 
-end # Transrate
+end # Transrate

data/lib/transrate/comparative_metrics.rb

@@ -4,7 +4,8 @@ module Transrate
   
   class ComparativeMetrics
 
-    attr_accessor :reciprocal_hits
+    attr_reader :rbh_per_contig
+    attr_reader :reciprocal_hits
 
     def initialize assembly, reference
       @assembly = assembly
@@ -17,8 +18,13 @@ module Transrate
       @ortholog_hit_ratio = self.ortholog_hit_ratio rbu
       @collapse_factor = self.collapse_factor @ra.r2l_hits
       @reciprocal_hits = rbu.size
+      @rbh_per_contig = @reciprocal_hits.to_f / @assembly.assembly.size.to_f
+    end
+
+    def comp_stats
       {
         :reciprocal_hits => @reciprocal_hits,
+        :rbh_per_contig => @rbh_per_contig,
         :ortholog_hit_ratio => @ortholog_hit_ratio,
         :collapse_factor => @collapse_factor
       }
@@ -44,8 +50,7 @@ module Transrate
         end
         targets[target] << query
       end
-      sum = targets.values.reduce(0.0){ |sum, val| sum += val.size }
-      puts "sum: #{sum}"
+      sum = targets.values.reduce(0.0){ |summer, val| summer += val.size }
       sum / targets.size
     end
 

data/lib/transrate/dimension_reduce.rb

@@ -0,0 +1,18 @@
+module Transrate
+
+  class DimensionReduce
+
+    def self.dimension_reduce(metrics)
+      total = 0
+      metrics.each do |metric|
+        o = metric.origin
+        w = metric.weighting
+        a = metric.score
+        total += w * ((o - a) ** 2)
+      end
+      Math.sqrt(total) / metrics.length
+    end
+      
+  end # DimensionReduce
+
+end # Transrate

data/lib/transrate/express.rb

@@ -0,0 +1,37 @@
+module Transrate
+
+  require 'which'
+  
+  class Express
+
+    # return an Express object
+    def initialize
+      express_path = Which::which('express')
+      raise "could not find eXpress in the path" if express_path.empty?
+      @express = express_path.first
+    end
+
+    # return hash of expression for each sequenceID
+    # in the assembly fastafile
+    def quantify_expression assembly, samfile
+      assembly = assembly.file if assembly.is_a? Assembly
+      cmd = "#{@express} --no-bias-correct #{assembly} #{samfile}"
+      ex_output = 'results.xprs'
+      fin_output = "#{assembly}_#{ex_output}"
+      unless File.exists? fin_output
+        `#{cmd}`
+        File.rename(ex_output, fin_output)
+      end
+      expression = {}
+      File.open(fin_output).each do |line|
+        line = line.chomp.split("\t")
+        target = line[1]
+        effective_count = line[7]
+        expression[target] = effective_count
+      end
+      expression
+    end
+    
+  end # Express
+
+ end # Transrate

data/lib/transrate/metric.rb

@@ -0,0 +1,16 @@
+module Transrate
+
+  class Metric
+
+    attr_reader :origin, :score, :name, :weighting
+    
+    def initialize(name, score, origin)
+      @origin = origin
+      @score = score
+      @name = name
+      @weighting = 1
+    end
+
+  end # Metric
+
+end # Transrate

data/lib/transrate/read_metrics.rb

@@ -5,7 +5,10 @@ module Transrate
     attr_reader :total
     attr_reader :bad
     attr_reader :supported_bridges
-
+    attr_reader :pc_good_mapping
+    attr_reader :percent_mapping
+    attr_reader :expressed_contigs
+    
     def initialize assembly
       @assembly = assembly
       @mapper = Bowtie2.new
@@ -14,12 +17,14 @@ module Transrate
 
     def run left, right, insertsize=200, insertsd=50
       @mapper.build_index @assembly.file
-      samfile = @mapper.map_reads(@assembly.file, 
-                                  left, right, 
-                                  insertsize, insertsd)
+      samfile = @mapper.map_reads(@assembly.file, left, right,  insertsize, insertsd)
       self.analyse_read_mappings(samfile, insertsize, insertsd)
+      self.analyse_expression(samfile)
       @percent_mapping = @total.to_f / @num_pairs.to_f * 100.0
       @pc_good_mapping = @good.to_f / @num_pairs.to_f * 100.0
+    end
+
+    def read_stats
       {
         :num_pairs => @num_pairs,
         :total_mappings => @total,
@@ -37,7 +42,9 @@ module Transrate
         :unrealistic_overlap => @unrealistic_overlap,
         :realistic_fragment => @realistic_fragment,
         :unrealistic_fragment => @unrealistic_fragment,
-        :potential_bridges => @supported_bridges
+        :potential_bridges => @supported_bridges,
+        :expressed_contigs => @expressed_contigs,
+        :unexpressed_contigs => @unexpressed_contigs
       }
     end
 
@@ -74,6 +81,8 @@ module Transrate
       @unrealistic_overlap = 0
       @realistic_fragment = 0
       @unrealistic_fragment = 0
+      @unexpressed_contigs = 0
+      @expressed_contigs = 0
     end
 
     def realistic_distance insertsize, insertsd
@@ -163,6 +172,19 @@ module Transrate
       end
     end
 
+    def analyse_expression samfile
+      express = Express.new
+      @expression = express.quantify_expression(@assembly.file, samfile)
+      @expression.each_pair do |target, count|
+        count = count.to_f
+        if count == 0
+          @unexpressed_contigs += 1
+        elsif count > 0
+          @expressed_contigs += 1
+        end
+      end
+    end
+    
   end # ReadMetrics
-
+  
 end # Transrate

data/lib/transrate/reciprocal_annotation.rb

@@ -22,10 +22,11 @@ module Transrate
 
     def make_assembly_db
       unless @assembly.orfs_ublast_db
+        assembly_dir = File.dirname(@assembly.file)
         assembly_base = File.basename(@assembly.file, ".*")
         assembly_orfs = assembly_base + ".orfs"
         @usearch.findorfs @assembly.file, assembly_orfs
-        assembly_db = assembly_base + ".udb"
+        assembly_db = File.join(assembly_dir, assembly_base + ".udb")
         @usearch.makeudb_ublast assembly_orfs, assembly_db
         @assembly.orfs_ublast_db = assembly_db
       end
@@ -33,8 +34,9 @@ module Transrate
 
     def make_reference_db
       unless @reference.ublast_db
+        reference_dir = File.dirname(@reference.file)
         reference_base = File.basename(@reference.file, ".*")
-        reference_db = reference_base + ".udb"
+        reference_db = File.join(reference_dir, reference_base + ".udb")
         @usearch.makeudb_ublast @reference.file, reference_db
         @reference.ublast_db = reference_db
       end
@@ -88,4 +90,4 @@ module Transrate
 
   end # ReciprocalAnnotation
 
-end # Transrate
+end # Transrate

data/lib/transrate/transrater.rb

@@ -7,11 +7,10 @@ module Transrate
     attr_reader :comparative_metrics
 
     def initialize assembly, reference, left, right, insertsize=nil, insertsd=nil
-      @assembly = Assembly.new assembly
+      @assembly  = assembly.is_a?(Assembly)  ? assembly  : Assembly.new(assembly)
       @reference = reference.is_a?(Assembly) ? reference : Assembly.new(reference)
       @read_metrics = ReadMetrics.new @assembly
       @comparative_metrics = ComparativeMetrics.new(@assembly, @reference)
-      self.run(left, right, insertsize, insertsd)
     end
 
     def run left, right, insertsize=nil, insertsd=nil
@@ -20,6 +19,22 @@ module Transrate
       @comparative_metrics.run
     end
 
+    def assembly_score
+      pg = Metric.new('pg', @read_metrics.pc_good_mapping, 0.0)
+      rbhpc = Metric.new('rbhpc', @comparative_metrics.rbh_per_contig, 0.0)
+      ec = Metric.new('ec', @read_metrics.expressed_contigs, 0.0)
+      @score = DimensionReduce.dimension_reduce([pg, rbhpc, ec])
+    end
+
+    def all_metrics left, right, insertsize=nil, insertsd=nil
+      self.run(left, right, insertsize, insertsd)
+      all = @assembly.basic_stats
+      all.merge!(@read_metrics.read_stats)
+      all.merge!(@comparative_metrics.comp_stats)
+      all[:score] = @score
+      all
+    end
+    
   end # Transrater
 
 end # Transrate

data/lib/transrate/usearch.rb

@@ -19,30 +19,36 @@ module Transrate
     end
 
     def ublast query, target, evalue="1e-5"
-      subcmd = " -ublast #{query}"
-      subcmd += " -db #{target}"
-      subcmd += " -evalue #{evalue}"
       blast6outfile = "#{File.basename(query)}_#{File.basename(target)}.b6"
-      subcmd += " -userout #{blast6outfile}"
-      subcmd += self.custom_output_fields
-      subcmd += " -strand both"
-      subcmd += " -threads #{@threads}"
-      self.run subcmd
+      unless File.exists? blast6outfile
+      subcmd = " -ublast #{query}"
+        subcmd += " -db #{target}"
+        subcmd += " -evalue #{evalue}"
+        subcmd += " -userout #{blast6outfile}"
+        subcmd += self.custom_output_fields
+        subcmd += " -strand both"
+        subcmd += " -threads #{@threads}"
+        self.run subcmd
+      end
       blast6outfile
     end
 
     def makeudb_ublast filepath, output
-      subcmd = " -makeudb_ublast #{filepath}"
-      subcmd += " -output #{output}"
-      self.run subcmd
+      unless File.exists? output
+        subcmd = " -makeudb_ublast #{filepath}"
+        subcmd += " -output #{output}"
+        self.run subcmd
+      end
     end
 
     def findorfs filepath, output
-      subcmd = " -findorfs #{filepath}"
-      subcmd += " -output #{output}"
-      subcmd += " -xlat"
-      subcmd += " -orfstyle 7"
-      self.run subcmd
+      unless File.exists? output
+        subcmd = " -findorfs #{filepath}"
+        subcmd += " -output #{output}"
+        subcmd += " -xlat"
+        subcmd += " -orfstyle 7"
+        self.run subcmd
+      end
     end
 
     def run subcmd
@@ -52,4 +58,4 @@ module Transrate
 
   end # Usearch
 
-end # Transrate
+end # Transrate

data/lib/transrate/version.rb

@@ -4,7 +4,7 @@ module Transrate
   module VERSION
     MAJOR = 0
     MINOR = 0
-    PATCH = 8
+    PATCH = 10
     BUILD = nil
 
     STRING = [MAJOR, MINOR, PATCH, BUILD].compact.join('.')

data/transrate.gemspec

@@ -7,7 +7,7 @@ Gem::Specification.new do |gem|
   gem.authors       = [ "Richard Smith" ]
   gem.email         = "rds45@cam.ac.uk"
   gem.licenses      = ["MIT"]
-  gem.homepage      = 'https://github.com/blahah/assemblotron'
+  gem.homepage      = 'https://github.com/blahah/transrate'
   gem.summary       = %q{ quality assessment of de-novo transcriptome assemblies }
   gem.description   = %q{ a library and command-line tool for quality assessment of de-novo transcriptome assemblies }
   gem.version       = Transrate::VERSION::STRING.dup

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: transrate
 version: !ruby/object:Gem::Version
-  version: 0.0.8
+  version: 0.0.10
 platform: ruby
 authors:
 - Richard Smith
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-09-25 00:00:00.000000000 Z
+date: 2013-09-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -150,9 +150,13 @@ files:
 - README.md
 - bin/transrate
 - lib/transrate.rb
+- lib/transrate/#assembly.rb#
 - lib/transrate/assembly.rb
 - lib/transrate/bowtie2.rb
 - lib/transrate/comparative_metrics.rb
+- lib/transrate/dimension_reduce.rb
+- lib/transrate/express.rb
+- lib/transrate/metric.rb
 - lib/transrate/rb_hit.rb
 - lib/transrate/read_metrics.rb
 - lib/transrate/reciprocal_annotation.rb
@@ -160,7 +164,7 @@ files:
 - lib/transrate/usearch.rb
 - lib/transrate/version.rb
 - transrate.gemspec
-homepage: https://github.com/blahah/assemblotron
+homepage: https://github.com/blahah/transrate
 licenses:
 - MIT
 metadata: {}