transrate 1.0.0.beta1 → 1.0.0.beta2

data/lib/transrate/read_metrics.rb

@@ -11,7 +11,7 @@ module Transrate
     def initialize assembly
       @assembly = assembly
       @mapper = Snap.new
-      @express = Express.new
+      @salmon = Salmon.new
       self.initial_values
 
       load_executables
@@ -19,7 +19,6 @@ module Transrate
     end
 
     def load_executables
-      @bam_splitter = get_bin_path 'bam-split'
       @bam_reader = get_bin_path 'bam-read'
     end
 
@@ -27,7 +26,7 @@ module Transrate
       which_bin = Cmd.new("which #{bin}")
       which_bin.run
       if !which_bin.status.success?
-        raise IOError.new("ReadMetrics: could not find #{bin} in path")
+        raise TransrateIOError.new("ReadMetrics: could not find #{bin} in path")
       end
       which_bin.stdout.split("\n").first
     end
@@ -35,10 +34,11 @@ module Transrate
     def run left, right, insertsize:200, insertsd:50, threads:8
       # check all read files exist
       [left, right].each do |readfile|
-        raise IOError.new "Read file is nil" if readfile.nil?
+        raise TransrateIOError.new "Read file is nil" if readfile.nil?
         readfile.split(",").each do |file|
           unless File.exist? file
-            raise IOError.new "ReadMetrics: read file does not exist: #{file}"
+            raise TransrateIOError.new "ReadMetrics: read file does not " +
+                                       "exist: #{file}"
           end
         end
       end
@@ -54,38 +54,18 @@ module Transrate
                                   threads: threads)
       @fragments = @mapper.read_count
 
-      sorted_bam = "#{File.basename(bamfile, '.bam')}.merged.sorted.bam"
-      merged_bam = "#{File.basename(bamfile, '.bam')}.merged.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"
+      assigned_bam = "postSample.bam"
+      final_bam = "#{File.basename(bamfile, '.bam')}.assigned.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
+      if !File.exist?(final_bam)
+        if !File.exist?(assigned_bam)
+          assigned_bam = assign_and_quantify bamfile
         end
-        sorted_bam = Samtools.sort_bam(merged_bam, [4, threads].min)
-        File.delete merged_bam
+        File.rename(assigned_bam, final_bam)
       end
-
       # analyse the final mappings
-      analyse_read_mappings(sorted_bam, insertsize, insertsd, true)
+      analyse_read_mappings(final_bam, insertsize, insertsd, true)
 
       @has_run = true
     end
@@ -133,31 +113,12 @@ module Transrate
       read_length
     end
 
-    def split_bam bamfile
-      base = File.basename(bamfile, '.bam')
-      valid = "#{base}.valid.bam"
-      invalid = "#{base}.invalid.bam"
-      if !File.exist? valid
-        cmd = "#{@bam_splitter} #{bamfile}"
-        splitter = Cmd.new cmd
-        splitter.run
-        if !splitter.status.success?
-          raise StandardError.new "Couldn't split bam file: #{bamfile}" +
-                      "\n#{splitter.stdout}\n#{splitter.stderr}"
-        end
-      end
-      if !File.exist? valid
-        raise StandardError.new "Splitting failed to create valid bam: #{valid}"
-      end
-      [valid, invalid]
-    end
-
     def assign_and_quantify bamfile
-      express_bam = @express.run(@assembly, bamfile)
+      @salmon.run(@assembly, bamfile)
     end
 
-    def analyse_expression express_output
-      express_output.each_pair do |name, expr|
+    def analyse_expression salmon_output
+      salmon_output.each_pair do |name, expr|
         contig_name = Bio::FastaDefline.new(name.to_s).entry_id
         contig = @assembly[contig_name]
         if expr[:eff_len]==0
@@ -190,14 +151,14 @@ module Transrate
         end
         @bad = @fragments_mapped - @good
       else
-        raise "couldn't find bamfile: #{bamfile}"
+        raise TransrateError.new "couldn't find bamfile: #{bamfile}"
       end
-      express_results = "#{File.basename @assembly.file}_results.xprs"
+      salmon_results = "#{File.basename @assembly.file}_quant.sf"
 
-      if File.exist?(express_results)
-        analyse_expression(@express.load_expression(express_results))
+      if File.exist?(salmon_results)
+        analyse_expression(@salmon.load_expression(salmon_results))
       else
-        abort "Can't find #{express_results}"
+        abort "Can't find #{salmon_results}"
       end
       @assembly.assembly.each_pair do |name, contig|
         @contigs_good += 1 if contig.score >= 0.5
@@ -228,7 +189,7 @@ module Transrate
         if !reader.status.success?
           msg = "Couldn't get information from bam file: #{bamfile}\n"
           msg << "#{reader.stdout}\n#{reader.stderr}"
-          raise msg
+          raise TransrateError.new msg
         end
       end
     end
@@ -236,8 +197,7 @@ module Transrate
     def populate_contig_data row
       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.p_seq_true = row[:p_seq_true]
       contig.uncovered_bases = row[:bases_uncovered]
       @bases_uncovered += contig.uncovered_bases
       if row[:fragments_mapped] and row[:fragments_mapped] > 1

data/lib/transrate/salmon.rb

@@ -0,0 +1,67 @@
+module Transrate
+
+  class SalmonError < TransrateError
+  end
+
+  class Salmon
+
+    def initialize
+      which = Cmd.new('which salmon')
+      which.run
+      if !which.status.success?
+        raise SalmonError.new("could not find salmon in the path")
+      end
+      @salmon = which.stdout.split("\n").first
+    end
+
+    def run assembly, bamfile, threads=8
+      assembly = assembly.file if assembly.is_a? Assembly
+      output = "quant.sf"
+      @fin_output = "#{File.basename assembly}_#{output}"
+      unless File.exist? @fin_output
+        salmon = Cmd.new build_command(assembly, bamfile, threads)
+        salmon.run
+        unless salmon.status.success?
+          logger.error salmon.stderr
+          raise SalmonError.new("Salmon failed")
+        end
+        File.rename(output, @fin_output)
+      end
+      return 'postSample.bam'
+    end
+
+    def build_command assembly, bamfile, threads=4
+      cmd = "#{@salmon} quant"
+      cmd << " --libType IU"
+      cmd << " --alignments #{bamfile}"
+      cmd << " --targets #{assembly}"
+      cmd << " --threads #{threads}"
+      cmd << " --sampleOut"
+      cmd << " --sampleUnaligned" # thanks Rob!
+      cmd << " --output ."
+      cmd
+    end
+
+    def load_expression file
+      expression = {}
+      File.open(file).each do |line|
+        if line !~ /^#/
+          line = line.chomp.split("\t")
+          target = line[0]
+          effective_length = line[1]
+          effective_count = line[4]
+          tpm = line[2]
+          expression[target] = {
+            :eff_len => effective_length.to_i,
+            :eff_count => effective_count.to_f,
+            :tpm => tpm.to_f
+          }
+        end
+      end
+      expression
+    end
+
+
+  end
+
+end

data/lib/transrate/snap.rb

@@ -1,6 +1,6 @@
 module Transrate
 
-  class SnapError < StandardError
+  class SnapError < TransrateError
   end
 
   class Snap
@@ -8,7 +8,7 @@ module Transrate
     require 'fix-trinity-output'
     require 'bio'
 
-    attr_reader :index_name, :sam, :read_count
+    attr_reader :index_name, :bam, :read_count
 
     def initialize
       which_snap = Cmd.new('which snap')
@@ -27,10 +27,6 @@ module Transrate
       l.split(",").zip(r.split(",")).each do |left, right|
         cmd << " #{left} #{right}"
       end
-      # NOTE: do NOT turn on the -so flag (sort bam output)
-      # it violates the basic assumption of eXpress's streaming
-      # algorithm: that the fragments are observed in approximately
-      # random order.
       cmd << " -o #{@bam}"
       cmd << " -s 0 1000" # min and max distance between paired-read starts
       cmd << " -H 300000" # max seed hits to consider in paired mode
@@ -39,7 +35,7 @@ module Transrate
       cmd << " -t #{threads}"
       cmd << " -b" # bind threads to cores
       cmd << " -M"  # format cigar string
-      cmd << " -D 5" # edit distance to search for mapq calculation
+      cmd << " -D 5" # extra edit distance to search. needed for -om
       cmd << " -om 5" # Output multiple alignments. extra edit distance
       cmd << " -omax 10" # max alignments per pair/read
       cmd
@@ -63,6 +59,7 @@ module Transrate
         save_readcount runner.stdout
         unless runner.status.success?
           if runner.stderr=~/Unmatched\sread\sIDs/
+            logger.warn runner.stderr
             logger.warn "Unmatched read IDs. Fixing input files..."
             remap_reads(left, right, threads)
           else
@@ -134,7 +131,6 @@ module Transrate
 
     def build_index file, threads
       @index_name = File.basename(file, File.extname(file))
-      file = check_ambiguous(file)
       unless Dir.exists?(@index_name)
         cmd = "#{@snap} index #{file} #{@index_name}"
         cmd << " -s 23"
@@ -151,30 +147,6 @@ module Transrate
       @index_built = true
     end
 
-    def check_ambiguous file
-
-      ref = Bio::FastaFormat.open(file)
-      ambiguous = false
-      fixed = ""
-      ref.each do |entry|
-        seq = entry.seq
-        if seq =~ /[RYSWKMBDHV]/
-          seq = seq.gsub(/[RYSWKMBDHV]/, "N")
-          ambiguous = true
-        end
-        fixed << ">#{entry.definition}\n#{seq}\n"
-      end
-      ref.close
-      if ambiguous
-        logger.warn "squelching ambiguous nucleotides"
-        file = "#{File.basename(file, File.extname(file))}.fixed.fasta"
-        File.open(file, "w") do |out|
-          out.write fixed
-        end
-      end
-      return file
-    end
-
   end # Snap
 
 end # Transrate

data/lib/transrate/transrater.rb

@@ -32,7 +32,7 @@ module Transrate
         end
         @read_metrics = ReadMetrics.new @assembly
       else
-        raise RuntimeError.new("assembly is nil")
+        raise TransrateError.new("assembly is nil")
       end
 
       if reference

data/lib/transrate/version.rb

@@ -11,7 +11,7 @@ module Transrate
     MAJOR = 1
     MINOR = 0
     PATCH = 0
-    BUILD = 'beta1'
+    BUILD = 'beta2'
 
     STRING = [MAJOR, MINOR, PATCH, BUILD].compact.join('.')
   end

data/lib/transrate.rb

@@ -3,25 +3,15 @@ require 'rbconfig'
 require 'yell'
 RbConfig::CONFIG['CFLAGS'] = ''
 
-require 'transrate/transrater'
-require 'transrate/version'
-require 'transrate/contig'
-require 'transrate/assembly'
-require 'transrate/snap'
-require 'transrate/score_optimiser'
-require 'transrate/express'
-require 'transrate/read_metrics'
-require 'transrate/comparative_metrics'
-require 'transrate/contig_metrics'
-require 'transrate/samtools'
-require 'transrate/cmd'
-require 'transrate/sam_checker'
-require 'transrate/transrate.so'
-
 # Transrate is a comprehensive transcriptome assembly
 # quality assessment tool.
 module Transrate
 
+  # Our own set of errors to allow nice custom error handling
+  class TransrateError < StandardError; end
+  class TransrateIOError < TransrateError; end
+  class TransrateArgError < TransrateError; end
+
   # Create the universal logger and include it in Object
   # making the logger object available everywhere
   format = Yell::Formatter.new("[%5L] %d : %m", "%Y-%m-%d %H:%M:%S")
@@ -35,3 +25,16 @@ module Transrate
   Object.send :include, Yell::Loggable
 
 end # Transrate
+
+require 'transrate/transrater'
+require 'transrate/version'
+require 'transrate/contig'
+require 'transrate/assembly'
+require 'transrate/snap'
+require 'transrate/score_optimiser'
+require 'transrate/salmon'
+require 'transrate/read_metrics'
+require 'transrate/comparative_metrics'
+require 'transrate/contig_metrics'
+require 'transrate/cmd'
+require 'transrate/transrate.so'