transrate 1.0.0.beta1 → 1.0.0.beta2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 015defaf8abd6b99791790ba1d59b016345c78b7
-  data.tar.gz: ea71a0e84c35c4fbc002314f8730530671c5b38c
+  metadata.gz: b90d296192b895be2cb4f5411ccc716264f5f360
+  data.tar.gz: 54fa1d66571912a1526aa7bd0d9ed549a04c8b19
 SHA512:
-  metadata.gz: 1c27bdcb6cdc2bf1c855fd8d30e1b834a09e1557974880b7f18a06f0eca7cd383504237a2f299bfa90e1f52bcbe3395b9b263d021c3ebc3bd02f092f6df9115d
-  data.tar.gz: bdcbe15c1dcd33233aada0b55955c8ad6d7cf1cc73f54893ea131f2e7f445b6447d9d2f47de8a1b82786413db275966d3e7db80d65e84a020020d2c22b74a3f1
+  metadata.gz: dc43e7f297ce2d53a03f9a7fcdde87124c605750cb6947d35e32185f5de92f53785fab37302e3abde3e4a2f7736aee3680fb415ab413d914720c025ebbff5f87
+  data.tar.gz: a122304dae2f6009c22e9d396c2b2f874cfa47937e582b1b5e5342cdf2a6d7eaeddcca465b3571022f7979d385f1c4014541a1ac1c5ca241eacaa8e8a4fd84e6

data/.gitignore

@@ -19,6 +19,7 @@ tmp
 \#*
 *so
 dryrun
+.DS_Store
 
 # YARD artifacts
 .yardoc

data/.travis.yml

@@ -2,8 +2,16 @@ language: ruby
 rvm:
   - "2.0.0"
   - "2.1.0"
+  - "2.2.0"
 before_script:
   - gem install bindeps
   - bundle install
   - bundle exec rake compile
   - bundle exec bin/transrate --install-deps
+os:
+  - linux
+  - osx
+matrix:
+  allow_failures:
+    - os: osx
+    - rvm: "2.2.0"

data/CITATION

@@ -0,0 +1,3 @@
+To cite transrate in publications, please use:
+
+Smith-Unna, Richard D.; Boursnell, Chris M.; Hibberd, Julian M. and Kelly, Steven (2014). Transrate: v1.0.0 beta 1. Github: https://github.com/Blahah/transrate. 10.5281/zenodo.13161

data/README.md

@@ -24,7 +24,7 @@ This software is being actively developed. Please be aware that there may be bug
 
 ## Citation
 
-Transrate is pre-publication academic software. If you use it, please cite the github repository and the DOI: [![DOI](https://zenodo.org/badge/3687/Blahah/transrate.png)](http://dx.doi.org/10.5281/zenodo.11039).
+Transrate is pre-publication academic software. If you use it, please cite the github repository and the DOI: [![DOI](https://zenodo.org/badge/doi/10.5281/zenodo.13161.svg)](http://dx.doi.org/10.5281/zenodo.13161).
 
 ## Documentation
 

data/Rakefile

@@ -57,5 +57,76 @@ Rake::TestTask.new do |t|
   t.test_files = ['test/test_contig.rb']
 end
 
+Rake::TestTask.new do |t|
+  t.name = :assembly
+  t.libs << 'test'
+  t.test_files = ['test/test_assembly.rb']
+end
+
+Rake::TestTask.new do |t|
+  t.name = :snap
+  t.libs << 'test'
+  t.test_files = ['test/test_snap.rb']
+end
+
+Rake::TestTask.new do |t|
+  t.name = :salmon
+  t.libs << 'test'
+  t.test_files = ['test/test_salmon.rb']
+end
+
+
 desc "Run tests"
 task :default => :test
+
+# PACKAGING
+
+PACKAGE_NAME = "transrate"
+VERSION = "1.0.0.beta2"
+TRAVELING_RUBY_VERSION = "20141215-2.1.5"
+
+desc "Package your app"
+task :package => ['package:linux:x86_64', 'package:osx']
+
+namespace :package do
+  namespace :linux do
+    desc "Package your app for Linux x86_64"
+    task :x86_64 => "packaging/packaging/traveling-ruby-#{TRAVELING_RUBY_VERSION}-linux-x86_64.tar.gz" do
+      create_package("linux-x86_64")
+    end
+  end
+
+  desc "Package your app for OS X"
+  task :osx => "packaging/packaging/traveling-ruby-#{TRAVELING_RUBY_VERSION}-osx.tar.gz" do
+    create_package("osx")
+  end
+end
+
+file "packaging/packaging/traveling-ruby-#{TRAVELING_RUBY_VERSION}-linux-x86_64.tar.gz" do
+  download_runtime("linux-x86_64")
+end
+
+file "packaging/packaging/traveling-ruby-#{TRAVELING_RUBY_VERSION}-osx.tar.gz" do
+  download_runtime("osx")
+end
+
+def create_package(target)
+  package_dir = "packaging/#{PACKAGE_NAME}-#{VERSION}-#{target}"
+  sh "rm -rf #{package_dir}"
+  sh "mkdir -p #{package_dir}/lib/app"
+  sh "cp -r lib #{package_dir}/lib/app/"
+  sh "cp -r bin #{package_dir}/lib/app/"
+  sh "mkdir #{package_dir}/lib/ruby"
+  sh "tar -xzf packaging/packaging/traveling-ruby-#{TRAVELING_RUBY_VERSION}-#{target}.tar.gz -C #{package_dir}/lib/ruby"
+  sh "cp packaging/transrate #{package_dir}/transrate"
+  if !ENV['DIR_ONLY']
+    sh "tar -czf #{package_dir}.tar.gz #{package_dir}"
+    sh "rm -rf #{package_dir}"
+  end
+end
+
+def download_runtime(target)
+  sh "mkdir -p packaging/packaging &&" +
+  "cd packaging/packaging && curl -L -O --fail " +
+  "http://d6r77u77i8pq3.cloudfront.net/releases/traveling-ruby-#{TRAVELING_RUBY_VERSION}-#{target}.tar.gz"
+end

data/bin/transrate

@@ -1,14 +1,32 @@
 #!/usr/bin/env ruby
-PROFILE = false
 
 require 'trollop'
 require 'transrate'
 require 'csv'
 require 'bindeps'
-require 'ruby-prof'
 
+include Transrate
+
+# Show the help message if no arguments provided
 ARGV[0] = "--help" if ARGV.length() == 0
 
+# We want clean error messages through the logger, no ugly backtraces
+# because the user doesn't care about them, unless they specifically ask for
+# them with --loglevel debug
+module Kernel
+  alias _raise raise
+
+  def raise(*a)
+    begin
+      _raise(*a)
+    rescue TransrateError => e
+      logger.error e.message
+      logger.debug e.backtrace unless e.backtrace.nil?
+      exit 1
+    end
+  end
+end
+
 opts = Trollop::options do
   version Transrate::VERSION::STRING.dup
   banner <<-EOS
@@ -19,8 +37,8 @@ opts = Trollop::options do
   DESCRIPTION:
   Analyse a de-novo transcriptome assembly using three kinds of metrics:
 
-  1. contig-based
-  2. read-mapping (if --left and --right are provided)
+  1. sequence-based (basic)
+  2. read-mapping-based (if --left and --right are provided)
   3. reference-based (if --reference is provided)
 
   Bug reports and feature requests at:
@@ -34,10 +52,10 @@ opts = Trollop::options do
   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
+  # basic assembly metrics only
   transrate --assembly contigs.fa
-  # contig and reference-based metrics with 8 threads
-  transrate --assembly contigs.fa --reference Athaliana_protein.fa --threads 8
+  # basic and reference-based metrics with 8 threads
+  transrate --assembly contigs.fa --reference Athaliana_transcripts.fa --threads 8
   # contig and read-based metrics for two assemblies with 32 threads
   transrate --assembly one.fa,two.fa --left l.fq --right r.fq --threads 32
 
@@ -61,61 +79,109 @@ opts = Trollop::options do
                  "one of [error, info, warn, debug]",
       :default => 'info'
   opt :install_deps, "install any missing dependencies"
+  opt :install_read_deps, "install missing dependencies for read metrics only"
+  opt :install_ref_deps, "install missing dependencies for reference metrics only"
 end
+
+# Check dependencies if they are relevant to the command issued,
+# and handle any commands to install missing ones
 gem_dir = Gem.loaded_specs['transrate'].full_gem_path
 gem_deps = File.join(gem_dir, 'deps', 'deps.yaml')
-if opts.install_deps
+blast_dep = File.join(gem_dir, 'deps', 'blast.yaml')
+
+if opts.install_deps || opts.install_read_deps || opts.install_ref_deps
+  # user has requested dependency installation
   puts "Checking dependencies"
-  Bindeps.require gem_deps
+
+  missing = []
+  if opts.install_deps || opts.install_read_deps
+    Bindeps.require gem_deps
+    missing += Bindeps.missing gem_deps
+  end
+
+  if opts.install_deps || opts.install_ref_deps
+    Bindeps.require blast_dep
+    missing += Bindeps.missing blast_dep
+  end
+
+  unless missing.empty?
+    raise TransrateError.new "Failed to install: \n - #{missing.join('\n - ')}"
+  end
+
   puts "All dependencies installed"
   exit
+
 else
-  missing = Bindeps.missing gem_deps
-  if missing.length > 0
+  # no dependency installation requested, but check dependencies
+  # for the commands provided are installed
+  missing = []
+  missing = Bindeps.missing gem_deps if opts.left
+  blast_missing = []
+  blast_missing = Bindeps.missing blast_dep if opts.reference
+
+  if missing.length + blast_missing.length > 0
     puts "Dependencies are missing:"
+
     missing.each do |dep|
       puts "  - #{dep.name} (#{dep.version})"
     end
-    puts "To install all missing dependencies, run `transrate --install-deps`"
-    exit(1)
+
+    blast_missing.each do |dep|
+      puts "  - #{dep.name} (#{dep.version})"
+    end
+
+    puts "To install all missing dependencies, run:"
+    puts "  transrate --install-deps"
+    puts "If you only want the read-metrics dependencies:"
+    puts "  transrate --install-read-deps"
+    puts "Or if you only want the reference-metrics dependencies: "
+    puts "  transrate --install-ref-deps"
+
+    exit 1
   end
+
+end
+
+# Handle commands
+unless %w[error info warn debug].include? opts.loglevel
+  raise "Loglevel #{opts.loglevel} is not valid. " +
+  "It must be one of: error, info, warn, debug."
 end
 
+logger.level = Yell::Level.new opts.loglevel.to_sym
+
 if opts.assembly
   opts.assembly.split(',').each do |assembly_file|
     unless File.exist?(assembly_file)
-      raise IOError.new "Assembly fasta file does not exist: #{assembly_file}"
+      raise TransrateIOError.new "Assembly fasta file does not exist: " +
+                                 " #{assembly_file}"
     end
   end
 else
-  raise ArgumentError.new "Option --assembly must be specified. " +
-                          "Try --help for help."
+  raise TransrateArgError.new "Option --assembly must be specified. " +
+                              "Try --help for help."
 end
 
 if opts.reference && !File.exist?(opts.reference)
-  raise IOError.new "Reference fasta file does not exist: #{opts.reference}"
+  raise TransrateIOError.new "Reference fasta file does not exist: " +
+                             " #{opts.reference}"
 end
 
 if opts.left and opts.right
   if opts.left.split(",").length != opts.right.split(",").length
     msg = "Please provide the same number of left reads as right reads"
-    raise ArgumentError.new(msg)
+    raise TransrateArgError.new msg
   end
   opts.left.split(",").zip(opts.right.split(",")).each do |left,right|
     if !File.exist?(left)
-      raise IOError.new "Left read fastq file does not exist: #{left}"
+      raise TransrateIOError.new "Left read fastq file does not exist: #{left}"
     end
     if !File.exist?(right)
-      raise IOError.new "Right read fastq file does not exist: #{right}"
+      raise TransrateIOError.new "Right read fastq file does not exist: #{right}"
     end
   end
 end
 
-if PROFILE
-  logger.info "Starting profiler"
-  RubyProf.start
-end
-
 def pretty_print_hash hash, width
   hash.map do |k, v|
     # show as float if there are any decimal places
@@ -133,15 +199,6 @@ def pretty_print_hash hash, width
   end
 end
 
-include Transrate
-
-unless %w[error info warn debug].include? opts.loglevel
-  raise "Loglevel #{opts.loglevel} is not valid. " +
-        "It must be one of: error, info, warn, debug."
-end
-
-logger.level = Yell::Level.new opts.loglevel.to_sym
-
 r = opts.reference ? Assembly.new(opts.reference) : nil
 report_width = 35
 
@@ -251,6 +308,7 @@ opts.assembly.split(',').each do |assembly|
   all << contig_results.merge(read_results)
                        .merge(comparative_results)
                        .merge({ :assembly => assembly })
+                       .merge({ :score => score })
 
 end
 
@@ -266,10 +324,3 @@ CSV.open(outfile, 'wb') do |file|
     file << head.map { |x| row[x] }
   end
 end
-
-if PROFILE
-  logger.info "Writing profiling results to transrate_profile.txt"
-  result = RubyProf.stop
-  printer = RubyProf::FlatPrinter.new(result)
-  printer.print(File.open('transrate_profile.txt', 'w'))
-end

data/deps/blast.yaml

@@ -0,0 +1,27 @@
+blastplus:
+  binaries:
+    - makeblastdb
+    - blastn
+    - tblastn
+    - blastp
+    - blastx
+    - tblastx
+    - makembindex
+    - psiblast
+    - rpsblast
+    - blastdbcmd
+    - segmasker
+    - dustmasker
+    - blast_formatter
+    - windowmasker
+    - blastdb_aliastool
+    - deltablast
+    - rpstblastn
+    - blastdbcheck
+  version:
+    number: '2.2.[0-9]'
+    command: 'blastx -version'
+  url:
+    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

data/deps/deps.yaml

@@ -1,80 +1,54 @@
-blastplus:
-  binaries:
-    - makeblastdb
-    - blastn
-    - tblastn
-    - blastp
-    - blastx
-    - tblastx
-    - makembindex
-    - psiblast
-    - rpsblast
-    - blastdbcmd
-    - segmasker
-    - dustmasker
-    - blast_formatter
-    - windowmasker
-    - blastdb_aliastool
-    - deltablast
-    - rpstblastn
-    - blastdbcheck
-  version:
-    number: '2.2.29'
-    command: 'blastx -version'
-  url:
-    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
 snap:
   binaries:
     - snap
   version:
-    number: '1.0dev.63'
+    number: '1.0dev.67.trfix1'
     command: 'snap'
   url:
     64bit:
-      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
-    - bcftools
-  version:
-    number: '0.1.19'
-    command: 'samtools'
-  url:
-    64bit:
-      linux: https://github.com/cboursnell/samtools/raw/master/build/linux64.tar.gz
-      macosx: https://github.com/cboursnell/samtools/raw/master/build/osx64.tar.gz
+      linux: https://github.com/Blahah/snap/releases/download/v1.0dev.67.trfix1/snap_v1.0dev.67.trfix1_linux.tar.gz
+      macosx: https://github.com/Blahah/snap/releases/download/v1.0dev.67.trfix1/snap_v1.0dev.67.trfix1_macosx.tar.gz
 bam-read:
   binaries:
    - bam-read
   version:
-    number: '1.0.0.beta1'
+    number: '1.0.0.beta3'
     command: 'bam-read'
   url:
     64bit:
-      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: '1.0.0.beta1'
-    command: 'bam-split'
-  url:
-    64bit:
-      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:
+      linux: https://github.com/Blahah/transrate-tools/releases/download/v1.0.0.beta3/bam-read_1.0.0.beta3_linux.tar.gz
+      macosx: https://github.com/Blahah/transrate-tools/releases/download/v1.0.0.beta3/bam-read_1.0.0.beta3_macosx.tar.gz
+  unpack: true
+salmon:
   binaries:
-    - express
+    - salmon
+  libraries:
+    - libbz2.so.1
+    - libgcc_s.so.1
+    - libgomp.so.1
+    - liblzma.so.0
+    - libm.so.6
+    - libpthread.so.0
+    - librt.so.1
+    - libstdc++.so.6
+    - libtbbmalloc_proxy.so,
+    - libtbbmalloc_proxy.so.2
+    - libtbbmalloc.so
+    - libtbbmalloc.so.2
+    - libtbb.so
+    - libtbb.so.2
+    - libz.so.1
+    - libcmph.0.dylib
+    - libcmph.dylib
+    - libcmph.la
+    - libstaden-read.la
+    - libtbb.dylib
+    - libtbbmalloc.dylib
+    - libtbbmalloc_proxy.dylib
   version:
-    number: '1.5.1'
-    command: 'express --version'
+    number: 'Salmon v0.2.7'
+    command: 'salmon --help'
   url:
     64bit:
-      linux: http://bio.math.berkeley.edu/eXpress/downloads/express-1.5.1/express-1.5.1-linux_x86_64.tgz
-      macosx: http://bio.math.berkeley.edu/eXpress/downloads/express-1.5.1/express-1.5.1-macosx_x86_64.tgz
+      linux: https://github.com/kingsfordgroup/sailfish/releases/download/v0.2.7/Salmon-v0.2.7_Ubuntu-12.04.tar.gz
+      macosx: https://github.com/kingsfordgroup/sailfish/releases/download/v0.2.7/Salmon-v0.2.7_MacOSX-10.10.1.tar.gz

data/ext/transrate/transrate.c

@@ -55,6 +55,9 @@ VALUE method_composition(VALUE self, VALUE _seq) {
   }
   for (i=0; i < len; i++) {
     base = seq[i];
+    if (base > 90) {
+      base -= 32;
+    }
     switch (base) {
       case 'A': {
         idx=0;
@@ -81,6 +84,9 @@ VALUE method_composition(VALUE self, VALUE _seq) {
 
     if (i > 0) {
       prevbase = seq[i-1];
+      if (prevbase > 90) {
+        prevbase -= 32;
+      }
       switch (prevbase) {
         case 'A': {
           idx=idx;
@@ -138,6 +144,9 @@ VALUE method_kmer_count(VALUE self, VALUE _k, VALUE _s) {
     n = 0;
     for(i = start; i < start+k; i++) {
       base = c_str[i];
+      if (base > 90) {
+        base -= 32;
+      }
       switch (base) {
         case 'A': {
           h = h << 2;

data/lib/transrate/assembly.rb

@@ -4,6 +4,8 @@ require 'forwardable'
 
 module Transrate
 
+  class AssemblyError < TransrateError; end
+
   # Container for a transcriptome assembly and its associated
   # metadata.
   #
@@ -41,13 +43,20 @@ module Transrate
     def initialize file
       @file = File.expand_path file
       unless File.exist? @file
-        raise IOError.new "Assembly file doesn't exist: #{@file}"
+        raise TransrateIOError.new "Assembly file doesn't exist: #{@file}"
       end
       @assembly = {}
       @n_bases = 0
       Bio::FastaFormat.open(file).each do |entry|
         @n_bases += entry.length
         contig = Contig.new(entry)
+        if @assembly.key?(contig.name)
+          logger.error "Non unique fasta identifier found"
+          logger.error ">#{contig.name}"
+          logger.error "Please make sure there are no duplicate entries in the assembly"
+          logger.error "Contig name is taken from before the first | or space"
+          raise AssemblyError
+        end
         @assembly[contig.name] = contig
       end
       @contig_metrics = ContigMetrics.new self
@@ -125,7 +134,6 @@ module Transrate
       # and iterate over them
       bin.sort_by! { |c| c.seq.length }
       bin.each do |contig|
-
         # increment our long contig counters if this
         # contig is above the thresholds
         if contig.length < 200
@@ -157,7 +165,6 @@ module Transrate
             cutoff = x2.pop / 100.0
           end
         end
-
       end
 
       # if there aren't enough sequences we might have no value for some
@@ -168,6 +175,11 @@ module Transrate
 
       # calculate and return the statistics as a hash
       mean = cumulative_length / @assembly.size
+      if @assembly.size * mean == 0
+        mean_orf_percent = 0
+      else
+        mean_orf_percent = 300 * orf_length_sum / (@assembly.size * mean)
+      end
       ns = Hash[x.map { |n| "n#{n}" }.zip(res)]
       {
         'n_seqs' => bin.size,
@@ -179,7 +191,7 @@ module Transrate
         'n_over_1k' => n_over_1k,
         'n_over_10k' => n_over_10k,
         'n_with_orf' => n_with_orf,
-        'mean_orf_percent' => 300 * orf_length_sum / (@assembly.size * mean)
+        'mean_orf_percent' => mean_orf_percent
       }.merge ns
 
     end # basic_bin_stats
@@ -187,20 +199,18 @@ module Transrate
     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
+      files = {}
+      %w(good fragmented chimeric bad).each do |type|
+        files[type.to_sym] = File.open("#{type}.#{base}", "wb")
       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
+        handle.write contig.to_fasta
       end
       # close all the file handles
-      files.each do |handle|
+      files.each do |type, handle|
         handle.close
       end
     end

data/lib/transrate/comparative_metrics.rb

@@ -55,12 +55,12 @@ module Transrate
       crbblast.reciprocals.each do |key, list|
         list.each_with_index do |hit, i|
           unless @reference.assembly.key? hit.target
-            raise "#{hit.target} not in reference"
+            raise TransrateError.new "#{hit.target} not in reference"
           end
           @reference[hit.target].hits << hit
 
           unless @assembly.assembly.key? hit.query
-            raise "#{hit.query} not in assembly"
+            raise TransrateError.new "#{hit.query} not in assembly"
           end
           contig = @assembly[hit.query]
           contig.has_crb = true

data/lib/transrate/contig.rb

@@ -20,7 +20,12 @@ module Transrate
     attr_accessor :hits
 
     def initialize(seq, name: nil)
-      seq.seq.gsub!("\0", "") # there is probably a better fix than this
+      # fix null bytes in the nucleotide sequence
+      seq.seq.gsub!("\0", "")
+      # trim trailing semicolons (because BLAST strips them)
+      if seq.respond_to?(:entry_id)
+        seq.entry_id.gsub!(/;$/, '')
+      end
       @seq = seq
       @seq.data = nil # no need to store raw fasta string
       @name = seq.respond_to?(:entry_id) ? seq.entry_id : name
@@ -248,13 +253,13 @@ module Transrate
       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
+        if p_not_segmented * p_bases_covered * p_seq_true >= 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
+        if p_good * p_bases_covered * p_seq_true >= 0.5
           return :chimeric
         end
       end