transrate 1.0.0.alpha.2 → 1.0.0.alpha.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 42c9c17a154a468377a5175f9937eb849c97e796
-  data.tar.gz: ddf0feca917a5d832ddf10143622ef121cd0cb73
+  metadata.gz: e36efc13482215880b40860b64a69253f3dc886b
+  data.tar.gz: 89f65043547cedb0552a7b620ea845c003f77637
 SHA512:
-  metadata.gz: 995833e4d14e1ceddebfd4c6f8f4c6a53fe1267a0ed53017844161803d575e7a3eaf17761315211b750fdf81c297179d17a975e737f1d2d17ca79e1a590f945e
-  data.tar.gz: 97f1ec6b36ec186e67c2c8b0550c6e90ac6ac4e59364a9c95c12b6c3ed516543a7143306ad641b8176c5da3148e0e762ddc03a42fc1c1236891832f5af73ba44
+  metadata.gz: 4e3ce4a80fc7bf7ef37f4a2322101f375b6efcf8bbed5dc8034f1bcbf871d2349292e2cc1a22f393bcabe3cddff60f4e1afd25d5267b1d3a2a7ea12e8a09ebf0
+  data.tar.gz: ee91ff26631d3204f43698a3afc4acfd0fc34a2068bde2c868897e347ff2bd8ca05225bc878267fc8454d58bd3a628fd7636735f9e8bd0472a7e3f90d1d829b9

data/bin/transrate

@@ -6,11 +6,14 @@ require 'csv'
 require 'bindeps'
 require 'ruby-prof'
 
+ARGV[0] = "--help" if ARGV.length() == 0
+
 opts = Trollop::options do
   version Transrate::VERSION::STRING.dup
   banner <<-EOS
 
-  Transrate v#{Transrate::VERSION::STRING.dup} by Richard Smith-Unna <rds45@cam.ac.uk>
+  Transrate v#{Transrate::VERSION::STRING.dup}
+  by Richard Smith-Unna <rds45@cam.ac.uk> and Chris Boursnell
 
   DESCRIPTION:
   Analyse a de-novo transcriptome assembly using three kinds of metrics:
@@ -26,8 +29,10 @@ opts = Trollop::options do
   transrate <options>
 
   EXAMPLES:
-  # just check dependencies and install any that are missing
+  # check dependencies and install any that are missing
   transrate --install-deps
+  # get the transrate score for the assembly and each contig
+  transrate --assembly contigs.fa --left left.fq --right right.fq
   # contig metrics only
   transrate --assembly contigs.fa
   # contig and reference-based metrics with 8 threads
@@ -57,7 +62,6 @@ opts = Trollop::options do
   opt :install_deps, "install any missing dependencies"
   opt :profile, "debug option: profile the code as it runs"
 end
-
 gem_dir = Gem.loaded_specs['transrate'].full_gem_path
 gem_deps = File.join(gem_dir, 'deps', 'deps.yaml')
 if opts.install_deps
@@ -70,7 +74,7 @@ else
   if missing.length > 0
     puts "Dependencies are missing:"
     missing.each do |dep|
-      puts "  - #{dep}"
+      puts "  - #{dep.name} (#{dep.version})"
     end
     puts "To install all missing dependencies, run `transrate --install-deps`"
     exit(1)

data/deps/deps.yaml

@@ -50,7 +50,7 @@ bam-read:
   binaries:
    - bam-read
   version:
-    number: '0.3\.1'
+    number: '0.3.2'
     command: 'bam-read'
   url:
     64bit:

data/lib/transrate.rb

@@ -22,7 +22,9 @@ module Transrate
 
   # Create the universal logger and include it in Object
   # making the logger object available everywhere
-  Yell.new(:format => "[%5L]: %m") do |l|
+  format = Yell::Formatter.new("[%5L] %d : %m", "%Y-%m-%d %H:%M:%S")
+  # http://xkcd.com/1179/
+  Yell.new(:format => format) do |l|
     l.level = :info
     l.name = Object
     l.adapter STDOUT, level: [:debug, :info, :warn]

data/lib/transrate/contig.rb

@@ -10,6 +10,7 @@ module Transrate
     def_delegators :@seq, :size, :length
     attr_accessor :seq, :name
     # read-based metrics
+    attr_accessor :eff_length, :eff_count, :tpm
     attr_accessor :coverage, :uncovered_bases, :p_uncovered_bases
     attr_accessor :p_seq_true, :p_unique
     attr_accessor :low_uniqueness_bases, :in_bridges
@@ -64,7 +65,10 @@ module Transrate
         :score => score,
         :p_unique => p_unique,
         :p_not_segmented => p_not_segmented,
-        :expression => coverage
+        :eff_length => eff_length,
+        :eff_count => eff_count,
+        :tpm => tpm,
+        :coverage => coverage
       }
     end
 

data/lib/transrate/express.rb

@@ -68,8 +68,14 @@ module Transrate
         end
         line = line.chomp.split("\t")
         target = line[1]
+        effective_length = line[3]
         effective_count = line[7]
-        expression[target] = effective_count.to_f
+        tpm = line[14]
+        expression[target] = {
+          :eff_len => effective_length.to_i,
+          :eff_count => effective_count.to_f,
+          :tpm => tpm.to_f
+        }
       end
       expression
     end

data/lib/transrate/read_metrics.rb

@@ -57,24 +57,30 @@ module Transrate
       # classify bam file into valid and invalid alignments
       sorted_bam = "#{File.basename(bamfile, '.bam')}.merged.sorted.bam"
       readsorted_bam = "#{File.basename(bamfile, '.bam')}.valid.sorted.bam"
-      unless File.exist? readsorted_bam
-        valid_bam, invalid_bam = split_bam bamfile
-        readsorted_bam = Samtools.readsort_bam(valid_bam, threads)
-        File.delete valid_bam
+      merged_bam = "#{File.basename(bamfile, '.bam')}.merged.bam"
+      unless File.exist? sorted_bam
+        unless File.exist? readsorted_bam
+          valid_bam, invalid_bam = split_bam bamfile
+          readsorted_bam = Samtools.readsort_bam valid_bam
+          File.delete valid_bam
+        end
       end
 
       # pass valid alignments to eXpress for assignment
       # always have to run the eXpress command to load the results
       assigned_bam = assign_and_quantify readsorted_bam
+      File.delete readsorted_bam if File.exist? readsorted_bam
 
       # merge the assigned alignments back with the invalid ones
       unless File.exist? sorted_bam
-        File.delete readsorted_bam
-        merged_bam = "#{File.basename(bamfile, '.bam')}.merged.bam"
-        Samtools.merge_bam(invalid_bam, assigned_bam, merged_bam, threads=threads)
-        File.delete invalid_bam
-        File.delete assigned_bam
-        sorted_bam = Samtools.sort_bam(merged_bam, threads)
+        unless File.exist? merged_bam
+          Samtools.merge_bam(invalid_bam, assigned_bam,
+                             merged_bam, threads=threads)
+
+          File.delete invalid_bam
+          File.delete assigned_bam
+        end
+        sorted_bam = Samtools.sort_bam(merged_bam, [4, threads].min)
         File.delete merged_bam
       end
 
@@ -154,11 +160,15 @@ module Transrate
     end
 
     def analyse_expression express_output
-      express_output.each_pair do |name, eff_count|
-        @contigs_uncovered += 1 if eff_count < 1
-        @contigs_lowcovered += 1 if eff_count < 10
+      express_output.each_pair do |name, expr|
         contig = @assembly[name]
-        contig.coverage = eff_count
+        coverage = expr[:eff_count] * @read_length / expr[:eff_len]
+        @contigs_uncovered += 1 if coverage < 1
+        @contigs_lowcovered += 1 if coverage < 10
+        contig.coverage = coverage.round(2)
+        contig.eff_length = expr[:eff_len]
+        contig.eff_count = expr[:eff_count]
+        contig.tpm = expr[:tpm]
       end
     end
 

data/lib/transrate/samtools.rb

@@ -47,7 +47,7 @@ module Transrate
       sorted = File.basename(bamfile, '.bam') + '.sorted'
       if !File.exist?("#{sorted}.bam")
         cmd = "sort"
-        cmd << " -l #{threads}"
+        cmd << " -@ #{threads}"
         cmd << " #{File.expand_path bamfile} #{sorted}"
         Samtools.run cmd
       end
@@ -63,7 +63,7 @@ module Transrate
       sorted = File.basename(bamfile, '.bam') + '.sorted'
       if !File.exist?("#{sorted}.bam")
         cmd = "sort"
-        cmd << " -l #{threads}"
+        cmd << " -@ #{threads}"
         cmd << " -n" # sort by read name only
         cmd << " #{File.expand_path bamfile} #{sorted}"
         Samtools.run cmd

data/lib/transrate/version.rb

@@ -11,7 +11,7 @@ module Transrate
     MAJOR = 1
     MINOR = 0
     PATCH = 0
-    BUILD = 'alpha.2'
+    BUILD = 'alpha.3'
 
     STRING = [MAJOR, MINOR, PATCH, BUILD].compact.join('.')
   end

data/test/test_bin.rb

@@ -93,6 +93,10 @@ class TestTransrateBin < Test::Unit::TestCase
       cmd << " --right #{right.join(",")}"
       c = Transrate::Cmd.new("#{cmd}")
       c.run
+      if !(c.status.success?)
+        puts c.stdout
+        puts c.stderr
+      end
       assert_equal true, c.status.success?, "exit status"
       assert File.exist?("transrate_assemblies.csv")
       assert File.exist?("transrate_sorghum_transcript.fa_contigs.csv"),

data/test/test_read_metrics.rb

@@ -73,12 +73,12 @@ class TestReadMetrics < Test::Unit::TestCase
           unc_a = contigs[0].uncovered_bases
           unc_b = contigs[1].uncovered_bases
           assert_equal 11, unc_a, "uncovered bases"
-          assert_equal 1, unc_b, "uncovered bases"
+          assert_equal 2, unc_b, "uncovered bases"
 
           prop_unc_a = a[:p_bases_covered]
           prop_unc_b = b[:p_bases_covered]
           assert_equal 0.98497, prop_unc_a.round(5), "proportion covered bases"
-          assert_equal 0.99878, prop_unc_b.round(5), "proportion covered bases"
+          assert_equal 0.99757, prop_unc_b.round(5), "proportion covered bases"
 
         end
       end
@@ -100,7 +100,7 @@ class TestReadMetrics < Test::Unit::TestCase
           edit_a = a[:p_not_segmented].round(5)
           edit_b = b[:p_not_segmented].round(5)
           assert_equal 0.11444, edit_a, "probability not segmented 1"
-          assert_equal 0.79629, edit_b, "probability not segmented 2"
+          assert_equal 0.82779, edit_b, "probability not segmented 2"
 
         end
       end

data/test/test_transrater.rb

@@ -59,7 +59,7 @@ class TestTransrater < Test::Unit::TestCase
         Dir.chdir tmpdir do
           all = @rater.all_metrics(@left, @right)
           score = @rater.assembly_score
-          assert_equal 0.40755, score.round(5) # regression test
+          assert_equal 0.40908, score.round(5) # regression test
         end
       end
     end

data/transrate.gemspec

@@ -19,7 +19,7 @@ Gem::Specification.new do |gem|
 
   gem.add_dependency 'yell', '~> 2.0', '>= 2.0.4'
   gem.add_dependency 'trollop', '~> 2.0'
-  gem.add_dependency 'bindeps', '~> 0.1', '>= 0.1.2'
+  gem.add_dependency 'bindeps', '~> 0.1', '>= 0.1.3'
   gem.add_dependency 'bio', '~> 1.4', '>= 1.4.3'
   gem.add_dependency 'crb-blast', '~> 0.4', '>= 0.4.2'
   gem.add_dependency 'ruby-prof', '~> 0.15', '>= 0.15.1'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: transrate
 version: !ruby/object:Gem::Version
-  version: 1.0.0.alpha.2
+  version: 1.0.0.alpha.3
 platform: ruby
 authors:
 - Richard Smith-Unna
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-10-17 00:00:00.000000000 Z
+date: 2014-10-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: yell
@@ -54,7 +54,7 @@ dependencies:
         version: '0.1'
     - - ">="
       - !ruby/object:Gem::Version
-        version: 0.1.2
+        version: 0.1.3
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
@@ -64,7 +64,7 @@ dependencies:
         version: '0.1'
     - - ">="
       - !ruby/object:Gem::Version
-        version: 0.1.2
+        version: 0.1.3
 - !ruby/object:Gem::Dependency
   name: bio
   requirement: !ruby/object:Gem::Requirement