transrate 1.0.0.alpha.8.bowtie → 1.0.0.beta1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0f6903da751cadec5102d2fa7e9c2381a9c39a25
-  data.tar.gz: bf9b30c2e334ae98ad6d959b085d521d43e5daf4
+  metadata.gz: 015defaf8abd6b99791790ba1d59b016345c78b7
+  data.tar.gz: ea71a0e84c35c4fbc002314f8730530671c5b38c
 SHA512:
-  metadata.gz: 97cdf2d6c922b1c0a901a1aa8f6e85013a8ec0c9cff54795b8bba3fa903efa7ef2ae4673c8e71e8ee292326eaadbf3ee617094c978a4fd39c836f6d8fb0d0672
-  data.tar.gz: 08ae932477980abe97b2b7fe874a14bd170688b1f8f60472971f21cc2ae0c7bb742d9a5c89ee923ccf5c476c4e59c5863a02a2e1994b4deb74e13e228e792136
+  metadata.gz: 1c27bdcb6cdc2bf1c855fd8d30e1b834a09e1557974880b7f18a06f0eca7cd383504237a2f299bfa90e1f52bcbe3395b9b263d021c3ebc3bd02f092f6df9115d
+  data.tar.gz: bdcbe15c1dcd33233aada0b55955c8ad6d7cf1cc73f54893ea131f2e7f445b6447d9d2f47de8a1b82786413db275966d3e7db80d65e84a020020d2c22b74a3f1

data/.gitignore

@@ -46,3 +46,6 @@ Makefile
 transrate.bundle
 transrate.o
 .RUBYARCHDIR.time
+
+# vagrant stuff
+.vagrant

data/README.md

@@ -2,6 +2,8 @@
   <img alt="Transrate - understand your transcriptome assembly" src="https://github.com/Blahah/transrate/raw/master/docs/transrate_logo_full.png">
 </p>
 
+Need help? Chat to us live: [![Gitter](https://badges.gitter.im/Join Chat.svg)](https://gitter.im/Blahah/transrate?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge)
+
 ## Development status
 
 [![Gem Version](http://img.shields.io/gem/v/transrate.svg)][gem]

data/bin/transrate

@@ -58,7 +58,7 @@ opts = Trollop::options do
   opt :outfile, "prefix filename to use for CSV output",
       :default => 'transrate'
   opt :loglevel, "the amount of information to print. " +
-                 "one of [error, warn, info, debug]",
+                 "one of [error, info, warn, debug]",
       :default => 'info'
   opt :install_deps, "install any missing dependencies"
 end
@@ -142,8 +142,6 @@ end
 
 logger.level = Yell::Level.new opts.loglevel.to_sym
 
-logger.info "Loading reference dataset"
-
 r = opts.reference ? Assembly.new(opts.reference) : nil
 report_width = 35
 
@@ -176,7 +174,7 @@ opts.assembly.split(',').each do |assembly|
     pretty_print_hash(contig_results, report_width)
   end
 
-  logger.info "Contig metrics done in #{(Time.now - t0).round(1)} seconds"
+  logger.info "Contig metrics done in #{Time.now - t0} seconds"
 
   read_results = {}
 
@@ -221,7 +219,7 @@ opts.assembly.split(',').each do |assembly|
   if (opts.left && opts.right)
     score = transrater.assembly_score
     unless score.nil?
-      logger.info "TRANSRATE ASSEMBLY SCORE: #{score.round(2)}"
+      logger.info "TRANSRATE ASSEMBLY SCORE: #{score.round(4)}"
       logger.info "-" * report_width
     end
   end

data/deps/deps.yaml

@@ -25,24 +25,16 @@ blastplus:
     64bit:
       macosx: ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.2.29/ncbi-blast-2.2.29+-universal-macosx.tar.gz
       linux: ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.2.29/ncbi-blast-2.2.29+-x64-linux.tar.gz
-bowtie2:
+snap:
   binaries:
-    - bowtie2
-    - bowtie2-align-l
-    - bowtie2-align-s
-    - bowtie2-build
-    - bowtie2-build-l
-    - bowtie2-build-s
-    - bowtie2-inspect
-    - bowtie2-inspect-l
-    - bowtie2-inspect-s
+    - snap
   version:
-    number: '2.2.3'
-    command: 'bowtie2 --version'
+    number: '1.0dev.63'
+    command: 'snap'
   url:
     64bit:
-      linux: http://downloads.sourceforge.net/project/bowtie-bio/bowtie2/2.2.3/bowtie2-2.2.3-linux-x86_64.zip
-      macosx: http://downloads.sourceforge.net/project/bowtie-bio/bowtie2/2.2.3/bowtie2-2.2.3-macos-x86_64.zip
+      linux: https://github.com/HibberdLab/snap/raw/dev/bin/linux/snap.tar.gz
+      macosx: https://github.com/HibberdLab/snap/raw/dev/bin/macosx/snap.tar.gz
 samtools:
   binaries:
     - samtools
@@ -58,23 +50,23 @@ bam-read:
   binaries:
    - bam-read
   version:
-    number: '0.3.4'
+    number: '1.0.0.beta1'
     command: 'bam-read'
   url:
     64bit:
-      linux: https://github.com/cboursnell/transrate-bam-read/raw/master/bin/linux/bam-read
-      macosx: https://github.com/Blahah/transrate-bam-read/raw/master/bin/macosx/bam-read
+      linux: https://github.com/cboursnell/transrate-tools/raw/master/bin/linux/bam-read
+      macosx: https://github.com/Blahah/transrate-tools/raw/master/bin/macosx/bam-read
   unpack: false
 bam-split:
   binaries:
    - bam-split
   version:
-    number: '0.2'
+    number: '1.0.0.beta1'
     command: 'bam-split'
   url:
     64bit:
-      linux: https://github.com/Blahah/transrate-bam-read/raw/master/bin/linux/bam-split
-      macosx: https://github.com/Blahah/transrate-bam-read/raw/master/bin/macosx/bam-split
+      linux: https://github.com/Blahah/transrate-tools/raw/master/bin/linux/bam-split
+      macosx: https://github.com/Blahah/transrate-tools/raw/master/bin/macosx/bam-split
   unpack: false
 express:
   binaries:

data/lib/transrate.rb

@@ -8,7 +8,7 @@ require 'transrate/version'
 require 'transrate/contig'
 require 'transrate/assembly'
 require 'transrate/snap'
-require 'transrate/bowtie'
+require 'transrate/score_optimiser'
 require 'transrate/express'
 require 'transrate/read_metrics'
 require 'transrate/comparative_metrics'

data/lib/transrate/assembly.rb

@@ -184,6 +184,27 @@ module Transrate
 
     end # basic_bin_stats
 
+    def classify_contigs
+      # create hash of file handles for each output
+      base = File.basename @file
+      files = Hash.new do
+        %w(good fragmented chimeric bad).each do |type|
+          handle = File.open("#{type}.#{base}", "wb")
+          [type.to_sym, handle]
+        end
+      end
+      # loop through contigs writing them out to the appropriate file
+      @assembly.each_pair do |name, contig|
+        category = contig.classify
+        handle = files[category]
+        handle.push contig.to_fasta
+      end
+      # close all the file handles
+      files.each do |handle|
+        handle.close
+      end
+    end
+
   end # Assembly
 
 end # Transrate

data/lib/transrate/contig.rb

@@ -12,8 +12,9 @@ module Transrate
     # 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, :in_bridges
-    attr_accessor :p_good, :p_not_segmented
+    attr_accessor :p_seq_true, :p_unique
+    attr_accessor :low_uniqueness_bases, :in_bridges
+    attr_accessor :p_good, :p_not_segmented, :good
     # reference-based metrics
     attr_accessor :has_crb, :reference_coverage
     attr_accessor :hits
@@ -34,6 +35,8 @@ module Transrate
       @p_unique = 0
       @p_not_segmented = 1
       @score = -1
+      @good = 0
+      @coverage = 0
     end
 
     def each &block
@@ -232,9 +235,36 @@ module Transrate
         [p_bases_covered, 0.01].max.to_f * # proportion of bases covered
         [p_not_segmented, 0.01].max.to_f * # prob contig has 0 changepoints
         [p_good, 0.01].max.to_f * # proportion of reads that mapped good
-        [p_seq_true, 0.01].max.to_f
+        [p_seq_true, 0.01].max.to_f # scaled 1 - mean per-base edit distance
       @score = [prod, 0.01].max
     end
+
+    # Classify the contig into one of the following classes:
+    # - good (score >= 0.5)
+    # - fragmented (in_bridges > 0) and no other problems
+    # - chimeric (p_not_segmented < 0.25) and no other problems
+    # - bad (score < 0.5 and not in any other category)
+    def classify
+      return :good if score >= 0.5
+      # fragmented?
+      if in_bridges > 5
+        if p_not_segmented * p_bases_covered * p_seq_true * p_unique >= 0.5
+          return :fragmented
+        end
+      end
+      # chimeric?
+      if p_not_segmented < 0.25
+        if p_good * p_bases_covered * p_seq_true * p_unique >= 0.5
+          return :chimeric
+        end
+      end
+      return :bad
+    end
+
+    def to_fasta
+      @seq.seq.to_fasta(@name)
+    end
+
   end
 
 end

data/lib/transrate/express.rb

@@ -8,6 +8,8 @@ module Transrate
 
     require 'ostruct'
 
+    attr_reader :fin_output
+
     # return an Express object
     def initialize
       which = Cmd.new('which express')
@@ -26,31 +28,34 @@ module Transrate
       assembly = assembly.file if assembly.is_a? Assembly
 
       ex_output = 'results.xprs'
-      fin_output = "#{File.basename assembly}_#{ex_output}"
+      @fin_output = "#{File.basename assembly}_#{ex_output}"
 
-      unless File.exists? fin_output
+      unless File.exists? @fin_output
         runner = Cmd.new build_command(assembly, bamfile)
         runner.run
         unless runner.status.success?
-          abort "express failed on the bam file\n#{runner.stderr}"
+          logger.warn "express failed. cleaning sam file and trying again"
+          File.delete("hits.1.samp.bam")
+          fix_problem_snap_output bamfile
+          runner.run
+          unless runner.status.success?
+            abort "express failed on the cleaned sam file\n#{runner.stderr}"
+          end
         end
-        File.rename(ex_output, fin_output)
+        File.rename(ex_output, @fin_output)
       end
-
-      OpenStruct.new(:results_file => fin_output,
-                     :expression => load_expression(fin_output),
-                     :align_samp => 'hits.1.samp.bam')
+      return 'hits.1.samp.bam'
     end
 
     # return the constructed eXpress command
     def build_command assembly, bamfile
       cmd = "#{@express}"
-      cmd << " #{File.expand_path assembly}"
-      cmd << " #{File.expand_path bamfile}"
       cmd << " --output-dir ."
       cmd << " --output-align-samp"
       cmd << " --no-update-check"
       cmd << " --additional-online 1"
+      cmd << " #{File.expand_path assembly}"
+      cmd << " #{File.expand_path bamfile}"
       cmd
     end
 
@@ -60,7 +65,7 @@ module Transrate
       expression = {}
       first = true
       File.open(file).each do |line|
-        if first
+        if first # skip header line
           first = false
           next
         end
@@ -78,19 +83,19 @@ module Transrate
       expression
     end
 
-    # def fix_problem_snap_output bam
-    #   # express failed, probably because of temporary snap error
-    #   # convert bam to sam
-    #   sam = "#{File.expand_path(File.basename(bam, File.extname(bam)))}.sam"
-    #   Samtools.run "view -h #{bam} > #{sam}"
-    #   # run sam fixer on sam
-    #   checker = SamChecker.new
-    #   fixed_sam = "#{File.expand_path(File.basename(sam, File.extname(sam)))}.fixed.sam"
-    #   checker.fix_sam(sam, fixed_sam)
-    #   # convert sam to bam
-    #   Samtools.run "view -bS #{fixed_sam} > #{bam}"
-    #   bam
-    # end
+    def fix_problem_snap_output bam
+      # express failed, probably because of temporary snap error
+      # convert bam to sam
+      sam = "#{File.expand_path(File.basename(bam, File.extname(bam)))}.sam"
+      Samtools.run "view -h #{bam} > #{sam}"
+      # run sam fixer on sam
+      checker = SamChecker.new
+      fixed_sam = "#{File.expand_path(File.basename(sam, File.extname(sam)))}.fixed.sam"
+      checker.fix_sam(sam, fixed_sam)
+      # convert sam to bam
+      Samtools.run "view -bS #{fixed_sam} > #{bam}"
+      bam
+    end
 
   end # Express
 

data/lib/transrate/read_metrics.rb

@@ -2,16 +2,16 @@ module Transrate
 
   class ReadMetrics
 
-    attr_reader :fragments_mapping
-    attr_reader :p_good_mapping
-    attr_reader :bad
+    attr_reader :fragments, :fragments_mapping, :p_good_mapping
+    attr_reader :good, :bad
     attr_reader :supported_bridges
     attr_reader :has_run
     attr_reader :read_length
 
     def initialize assembly
       @assembly = assembly
-      @mapper = Bowtie2.new
+      @mapper = Snap.new
+      @express = Express.new
       self.initial_values
 
       load_executables
@@ -47,31 +47,36 @@ module Transrate
       @read_length = get_read_length(left, right)
 
       # map reads
-      @mapper.build_index(@assembly.file)
-      samfile = @mapper.map_reads(@assembly.file, left, right,
+      @mapper.build_index(@assembly.file, threads)
+      bamfile = @mapper.map_reads(@assembly.file, left, right,
                                   insertsize: insertsize,
                                   insertsd: insertsd,
                                   threads: threads)
       @fragments = @mapper.read_count
-      bamfile = Samtools.sam_to_bam(samfile)
-      File.delete samfile
 
-      # classify bam file into valid and invalid alignments
       sorted_bam = "#{File.basename(bamfile, '.bam')}.merged.sorted.bam"
       merged_bam = "#{File.basename(bamfile, '.bam')}.merged.bam"
-      valid_bam, invalid_bam = split_bam bamfile
-      readsorted_bam = Samtools.readsort_bam valid_bam
+      assigned_bam = "hits.1.samp.bam"
+      readsorted_bam = "#{File.basename(bamfile, '.bam')}.readsorted.bam"
+      valid_bam = "#{File.basename(bamfile, '.bam')}.valid.bam"
+      invalid_bam = "#{File.basename(bamfile, '.bam')}.invalid.bam"
 
-      # 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
-
-      # merge the assigned alignments back with the invalid ones
-      unless File.exist? sorted_bam
-        unless File.exist? merged_bam
+      # check for latest files first and create what is needed
+      if !File.exist?(sorted_bam)
+        if !File.exist?(merged_bam)
+          if !File.exist?(assigned_bam)
+            if !File.exist?(readsorted_bam)
+              if !File.exist?(valid_bam)
+                valid_bam, invalid_bam = split_bam bamfile
+              end
+              readsorted_bam = Samtools.readsort_bam(valid_bam)
+              File.delete valid_bam
+            end
+            assigned_bam = assign_and_quantify readsorted_bam
+            File.delete readsorted_bam
+          end
           Samtools.merge_bam(invalid_bam, assigned_bam,
                              merged_bam, threads=threads)
-
           File.delete invalid_bam
           File.delete assigned_bam
         end
@@ -148,15 +153,13 @@ module Transrate
     end
 
     def assign_and_quantify bamfile
-      express = Express.new
-      results = express.run(@assembly, bamfile)
-      analyse_expression results.expression
-      results.align_samp
+      express_bam = @express.run(@assembly, bamfile)
     end
 
     def analyse_expression express_output
       express_output.each_pair do |name, expr|
-        contig = @assembly[name]
+        contig_name = Bio::FastaDefline.new(name.to_s).entry_id
+        contig = @assembly[contig_name]
         if expr[:eff_len]==0
           coverage = 0
         else
@@ -176,7 +179,7 @@ module Transrate
         csv_output = "#{File.basename(@assembly.file)}_bam_info.csv"
         csv_output = File.expand_path(csv_output)
 
-        analyse_bam(bamfile, csv_output)
+        analyse_bam bamfile, csv_output
         # open output csv file
         @potential_bridges = 0
 
@@ -189,6 +192,13 @@ module Transrate
       else
         raise "couldn't find bamfile: #{bamfile}"
       end
+      express_results = "#{File.basename @assembly.file}_results.xprs"
+
+      if File.exist?(express_results)
+        analyse_expression(@express.load_expression(express_results))
+      else
+        abort "Can't find #{express_results}"
+      end
       @assembly.assembly.each_pair do |name, contig|
         @contigs_good += 1 if contig.score >= 0.5
       end
@@ -212,7 +222,7 @@ module Transrate
 
     def analyse_bam bamfile, csv_output
       if !File.exist?(csv_output)
-        cmd = "#{@bam_reader} #{bamfile} #{csv_output}"
+        cmd = "#{@bam_reader} #{bamfile} #{csv_output} 0.7"
         reader = Cmd.new cmd
         reader.run
         if !reader.status.success?
@@ -224,12 +234,13 @@ module Transrate
     end
 
     def populate_contig_data row
-      contig = @assembly[row[:name]]
+      name = Bio::FastaDefline.new(row[:name].to_s).entry_id
+      contig = @assembly[name]
       scale = 0.7
       contig.p_seq_true = (row[:p_seq_true] - scale) * (1.0 / (1 - scale))
       contig.uncovered_bases = row[:bases_uncovered]
       @bases_uncovered += contig.uncovered_bases
-      if row[:fragments_mapped] and row[:fragments_mapped] > 0
+      if row[:fragments_mapped] and row[:fragments_mapped] > 1
         contig.p_good = row[:good]/row[:fragments_mapped].to_f
       end
       contig.p_not_segmented = row[:p_not_segmented]
@@ -242,6 +253,7 @@ module Transrate
         @potential_bridges += 1
       end
       @fragments_mapped += row[:fragments_mapped]
+      contig.good = row[:good]
       @good += row[:good]
       if row[:bases_uncovered] > 0
         @contigs_uncovbase += 1