genevalidator 1.6.12 → 2.1.3

data/lib/genevalidator/validation_length_cluster.rb

@@ -15,7 +15,9 @@ module GeneValidator
 
     def initialize(short_header, header, description, query_length, limits,
                    expected = :yes)
-      @short_header, @header, @description = short_header, header, description
+      @short_header = short_header
+      @header = header
+      @description = description
       @limits       = limits
       @query_length = query_length
       @expected     = expected
@@ -32,7 +34,7 @@ module GeneValidator
     end
 
     def explain
-      diff = (@result == :yes) ? 'inside' : 'outside'
+      diff = @result == :yes ? 'inside' : 'outside'
       'The most dense length-cluster of BLAST hits includes' \
       " sequences that are from #{@limits[0]} to #{@limits[1]} amino-acids" \
       " long. The query sequence is #{@query_length} amino-acids long and" \
@@ -44,7 +46,7 @@ module GeneValidator
         'There is no reason to believe there is any problem with the length' \
         ' of the query sequence.'
       else
-        size_diff  = (@query_length > @limits[1]) ? 'long' : 'short'
+        size_diff = @query_length > @limits[1] ? 'long' : 'short'
         "This suggests that the query sequence may be too #{size_diff}."
       end
     end
@@ -67,6 +69,9 @@ module GeneValidator
   # This class contains the methods necessary for
   # length validation by hit length clusterization
   class LengthClusterValidation < ValidationTest
+    extend Forwardable
+    def_delegators GeneValidator, :opt
+
     attr_reader :clusters
     attr_reader :max_density_cluster
 
@@ -96,8 +101,8 @@ module GeneValidator
     # Output:
     # +LengthClusterValidationOutput+ object
     def run
-      fail NotEnoughHitsError unless hits.length >= 5
-      fail unless prediction.is_a?(Query) && hits[0].is_a?(Query)
+      raise NotEnoughHitsError if hits.length < opt[:min_blast_hits]
+      raise unless prediction.is_a?(Query) && hits[0].is_a?(Query)
 
       start = Time.now
       # get [clusters, max_density_cluster_idx]
@@ -119,12 +124,11 @@ module GeneValidator
       @validation_report.run_time = Time.now - start
 
       @validation_report
-
     rescue NotEnoughHitsError
       @validation_report = ValidationReport.new('Not enough evidence', :warning,
                                                 @short_header, @header,
                                                 @description)
-    rescue
+    rescue StandardError
       @validation_report = ValidationReport.new('Unexpected error', :error,
                                                 @short_header, @header,
                                                 @description)
@@ -143,7 +147,7 @@ module GeneValidator
     def clusterization_by_length(_debug = false,
                                  lst = @hits,
                                  predicted_seq = @prediction)
-      fail TypeError unless lst[0].is_a?(Query) && predicted_seq.is_a?(Query)
+      raise TypeError unless lst[0].is_a?(Query) && predicted_seq.is_a?(Query)
 
       contents = lst.map { |x| x.length_protein.to_i }.sort { |a, b| a <=> b }
 
@@ -159,11 +163,10 @@ module GeneValidator
       end
 
       [clusters, max_density_cluster_idx]
-
     rescue TypeError => error
       error_location = error.backtrace[0].scan(%r{([^/]+:\d+):.*})[0][0]
-      $stderr.puts "Type error at #{error_location}."
-      $stderr.puts ' Possible cause: one of the arguments of the' \
+      warn "Type error at #{error_location}."
+      warn ' Possible cause: one of the arguments of the' \
                    ' "clusterization_by_length" method has not the proper type.'
       exit 1
     end
@@ -181,11 +184,11 @@ module GeneValidator
                             max_density_cluster = @max_density_cluster,
                             prediction = @prediction)
 
-      data = clusters.each_with_index.map { |cluster, i|
-        cluster.lengths.collect { |k, v|
+      data = clusters.each_with_index.map do |cluster, i|
+        cluster.lengths.collect do |k, v|
           { 'key' => k, 'value' => v, 'main' => (i == max_density_cluster) }
-        }
-      }
+        end
+      end
 
       Plot.new(data,
                :bars,

data/lib/genevalidator/validation_length_rank.rb

@@ -23,7 +23,9 @@ module GeneValidator
     def initialize(short_header, header, description, msg, query_length,
                    no_of_hits, median, mean, smallest_hit, largest_hit,
                    extreme_hits, percentage)
-      @short_header, @header, @description = short_header, header, description
+      @short_header = short_header
+      @header = header
+      @description = description
       @msg          = msg
       @query_length = query_length
       @no_of_hits   = no_of_hits
@@ -44,17 +46,17 @@ module GeneValidator
     end
 
     def explain
-      diff = (@query_length > @median) ? 'longer' : 'shorter'
+      diff = @query_length > @median ? 'longer' : 'shorter'
       exp1 = "The query sequence is  #{@query_length} amino-acids long. BLAST" \
              " identified #{@no_of_hits} hit sequences with lengths from" \
              " #{@smallest_hit} to #{@largest_hit} amino-acids (median:" \
              " #{@median}; mean: #{@mean})."
-      if @extreme_hits != 0
-        exp2 = " #{@extreme_hits} of these hit sequences (i.e." \
-               " #{@percentage}%) are #{diff} than the query sequence."
-      else
-        exp2 = " All hit sequences are #{diff} than the query sequence."
-      end
+      exp2 = if @extreme_hits != 0
+               " #{@extreme_hits} of these hit sequences (i.e." \
+                      " #{@percentage}%) are #{diff} than the query sequence."
+             else
+               " All hit sequences are #{diff} than the query sequence."
+             end
       exp1 + exp2
     end
 
@@ -68,11 +70,11 @@ module GeneValidator
     end
 
     def print
-      (@msg.empty?) ? "#{@percentage}%" : "#{@percentage}% (#{@msg})"
+      @msg.empty? ? "#{@percentage}%" : "#{@percentage}% (#{@msg})"
     end
 
     def validation
-      (@msg.empty?) ? :yes : :no
+      @msg.empty? ? :yes : :no
     end
   end
 
@@ -80,6 +82,9 @@ module GeneValidator
   # This class contains the methods necessary for
   # length validation by ranking the hit lengths
   class LengthRankValidation < ValidationTest
+    extend Forwardable
+    def_delegators GeneValidator, :opt
+
     THRESHOLD = 20
     ##
     # Initializes the object
@@ -104,13 +109,13 @@ module GeneValidator
     # Output:
     # +LengthRankValidationOutput+ object
     def run(hits = @hits, prediction = @prediction)
-      fail NotEnoughHitsError unless hits.length >= 5
-      fail unless prediction.is_a?(Query) && hits[0].is_a?(Query)
+      raise NotEnoughHitsError if hits.length < opt[:min_blast_hits]
+      raise unless prediction.is_a?(Query) && hits[0].is_a?(Query)
 
       start = Time.now
 
       hits_lengths = hits.map { |x| x.length_protein.to_i }
-                     .sort { |a, b| a <=> b }
+                         .sort { |a, b| a <=> b }
 
       no_of_hits   = hits_lengths.length
       median       = hits_lengths.median.round
@@ -147,12 +152,11 @@ module GeneValidator
                                                           percentage)
       @validation_report.run_time = Time.now - start
       @validation_report
-
     rescue NotEnoughHitsError
       @validation_report = ValidationReport.new('Not enough evidence', :warning,
                                                 @short_header, @header,
                                                 @description)
-    rescue
+    rescue StandardError
       @validation_report = ValidationReport.new('Unexpected error', :error,
                                                 @short_header, @header,
                                                 @description)

data/lib/genevalidator/validation_maker_qi.rb

@@ -10,7 +10,9 @@ module GeneValidator
   class MakerQIValidationOutput < ValidationReport
     def initialize(short_header, header, description, splice_sites, exons,
                    expected = :yes)
-      @short_header, @header, @description = short_header, header, description
+      @short_header = short_header
+      @header = header
+      @description = description
       @splice_sites = splice_sites
       @exons        = exons
       @expected     = expected
@@ -27,12 +29,12 @@ module GeneValidator
     def explain
       if @splice_sites == -100
         "#{@exons}% of exons match an EST/mRNA-seq alignment." \
-        " No splice sites were identified and as such cannot be confirmed by" \
-        " an EST/mRNA-seq alignment."
+        ' No splice sites were identified and as such cannot be confirmed by' \
+        ' an EST/mRNA-seq alignment.'
       else
         "#{@exons}% of exons match an EST/mRNA-seq alignment and" \
         " #{@splice_sites}% of splice sites are confirmed by EST/mRNA-seq" \
-        " alignment."
+        ' alignment.'
       end
     end
 
@@ -54,7 +56,7 @@ module GeneValidator
     end
 
     def validation
-      (@splice_sites > 80 && @exons > 80) ? :yes : :no
+      @splice_sites > 80 && @exons > 80 ? :yes : :no
     end
   end
 
@@ -77,7 +79,7 @@ module GeneValidator
     # Output:
     # +QIValidationOutput+ object
     def run
-      fail unless prediction.is_a?(Query)
+      raise unless prediction.is_a?(Query)
 
       start  = Time.now
 
@@ -87,7 +89,7 @@ module GeneValidator
                                              #{number}\|#{number}\|#{number}\|
                                              #{number}\|#{number}/x)
 
-      fail NotEnoughEvidence if match.nil?
+      raise NotEnoughEvidence if match.nil?
 
       # % of splice sites confirmed by EST/mRNA-seq alignment
       splice_sites = (match[1].to_f * 100).round
@@ -99,12 +101,11 @@ module GeneValidator
                                                        splice_sites, exons)
       @validation_report.run_time = Time.now - start
       @validation_report
-
     rescue NotEnoughEvidence
-      @validation_report =  ValidationReport.new('No MAKER Quality Index',
-                                                 :warning, @short_header,
-                                                 @header, @description)
-    rescue
+      @validation_report = ValidationReport.new('No MAKER Quality Index',
+                                                :warning, @short_header,
+                                                @header, @description)
+    rescue StandardError
       @validation_report = ValidationReport.new('Unexpected error', :error,
                                                 @short_header, @header,
                                                 @description)

data/lib/genevalidator/validation_open_reading_frame.rb

@@ -17,7 +17,9 @@ module GeneValidator
 
     def initialize(short_header, header, description, orfs, coverage,
                    longest_orf_frame, threshold = 80, expected = :yes)
-      @short_header, @header, @description = short_header, header, description
+      @short_header = short_header
+      @header = header
+      @description = description
 
       @orfs         = orfs
       @coverage     = coverage
@@ -28,9 +30,10 @@ module GeneValidator
       @mainORFFrame = longest_orf_frame
       @approach     = 'We expect the query sequence to encode a single gene,' \
                       ' thus it should contain one main Open Reading Frame' \
-                      ' (ORF) that occupies most of the query sequence.'
+                      ' (ORF) that occupies most of the query sequence.' \
+                      ' This validation does not require any BLAST hits.'
       @explanation  = " The longest ORF is in frame #{@mainORFFrame}, where" \
-                      " it occupies #{(@coverage).round}% of the query" \
+                      " it occupies #{@coverage.round}% of the query" \
                       ' sequence.'
       @conclusion   = conclude
     end
@@ -50,11 +53,11 @@ module GeneValidator
       orf_list = ''
       @orfs.map { |elem| orf_list << "#{elem[0]}:#{elem[1]}," }
 
-      "#{(@coverage).round}%&nbsp;(frame&nbsp;#{@mainORFFrame})"
+      "#{@coverage.round}%&nbsp;(frame&nbsp;#{@mainORFFrame})"
     end
 
     def validation
-      (@coverage > @threshold) ? :yes : :no
+      @coverage > @threshold ? :yes : :no
     end
   end
 
@@ -75,8 +78,8 @@ module GeneValidator
       @short_header = 'MainORF'
       @header       = 'Main ORF'
       @description  = 'Check whether there is a single main Open Reading' \
-                      ' Frame in the predicted gene. Applicable only for' \
-                      ' nucleotide queries.'
+                      ' Frame in the predicted gene. This validation does not' \
+                      ' require any BLAST hits.'
       @cli_name     = 'orf'
       @type         = config[:type]
     end
@@ -91,12 +94,12 @@ module GeneValidator
         return @validation_report
       end
 
-      fail unless prediction.is_a?(Query)
+      raise unless prediction.is_a?(Query)
 
       start = Time.new
       orfs = get_orfs
 
-      longest_orf       = orfs.sort_by { |_key, hash| hash[:coverage] }.last
+      longest_orf       = orfs.max_by { |_key, hash| hash[:coverage] }
       longest_orf_frame = longest_orf[1][:frame]
       coverage          = longest_orf[1][:coverage]
       translated_length = longest_orf[1][:translated_length]
@@ -109,7 +112,7 @@ module GeneValidator
 
       @validation_report.plot_files.push(plot1)
       @validation_report
-    rescue
+    rescue StandardError
       @validation_report = ValidationReport.new('Unexpected error', :error,
                                                 @short_header, @header,
                                                 @description)
@@ -137,8 +140,8 @@ module GeneValidator
         f = -2 if f == 5
         f = -3 if f == 6
         s.scan(/(\w{30,})/) do |_orf|
-          orf_start = $~.offset(0)[0] + 1
-          orf_end   = $~.offset(0)[1] + 1
+          orf_start = $LAST_MATCH_INFO.offset(0)[0] + 1
+          orf_end   = $LAST_MATCH_INFO.offset(0)[1] + 1
           coverage = (((orf_end - orf_start) / s.length.to_f) * 100).ceil
           # reduce the orf_end and the translated length by 2% to increase the
           #   width between ORFs on the plot
@@ -161,7 +164,7 @@ module GeneValidator
     # +output+: location where the plot will be saved in jped file format
     # +prediction+: Sequence objects
     def plot_orfs(orfs, translated_length)
-      fail QueryError unless orfs.is_a? Hash
+      raise QueryError unless orfs.is_a? Hash
 
       data = []
 

data/lib/genevalidator/validation_report.rb

@@ -63,7 +63,7 @@ module GeneValidator
       elsif validation == :error || validation == :unapplicable
         'warning'
       else
-        (validation == :warning) ? 'warning' : 'danger'
+        validation == :warning ? 'warning' : 'danger'
       end
     end
   end

data/lib/genevalidator/validation_test.rb

@@ -36,7 +36,7 @@ module GeneValidator
     end
 
     def run
-      fail 'run method should be implemented by all classes that extend' \
+      raise 'run method should be implemented by all classes that extend' \
            ' ValidationTest'
     end
   end

data/lib/genevalidator/version.rb

@@ -1,3 +1,3 @@
 module GeneValidator
-  VERSION = '1.6.12'
+  VERSION = '2.1.3'.freeze
 end

data/test/overall.rb

@@ -53,4 +53,4 @@ def run(*cmds)
 end
 
 db = ENV['db'] || 'swissprot -remote'
-%w(data/protein_data.fasta data/mrna_data.fasta).each { |fa| test_with fa, db }
+%w[data/protein_data.fasta data/mrna_data.fasta].each { |fa| test_with fa, db }

data/test/test_all_validations.rb

@@ -38,38 +38,45 @@ module GeneValidator
         original_stdout = $stdout.clone
         $stdout.reopen(prot_xml_out, 'w')
 
-        FileUtils.rm_rf(prot_output_dir) rescue Errno::ENOENT
         opts = {
-          validations: %w(lenc lenr frame merge dup orf align),
+          validations: %w[lenc lenr frame merge dup orf align],
           db: database,
           num_threads: threads,
-          fast: false,
+          mafft_threads: 1,
           input_fasta_file: prot_input,
           blast_xml_file: prot_xml,
           raw_sequences: prot_raw,
-          test: true
+          output_dir: prot_output_dir,
+          output_formats: %w[html csv json stdout],
+          min_blast_hits: 5,
+          test: true,
+          hide_summary: true,
+          force_rewrite: true
         }
 
-        GeneValidator.init(opts, 1, false)
+        GeneValidator.init(opts)
         GeneValidator.run
         $stdout.reopen original_stdout
         $stdout.reopen(prot_tab_out, 'w')
 
-        FileUtils.rm_rf(prot_output_dir) rescue Errno::ENOENT
-
         opts1 = {
-          validations: %w(lenc lenr frame merge dup orf align),
+          validations: %w[lenc lenr frame merge dup orf align],
           db: database,
           num_threads: threads,
-          fast: false,
+          mafft_threads: 1,
           input_fasta_file: prot_input,
           blast_tabular_file: prot_tab,
           blast_tabular_options: tab_options,
           raw_sequences: prot_raw,
-          test: true
+          output_dir: prot_output_dir,
+          output_formats: %w[html csv json stdout],
+          min_blast_hits: 5,
+          test: true,
+          hide_summary: true,
+          force_rewrite: true
         }
 
-        GeneValidator.init(opts1, 1, false)
+        GeneValidator.init(opts1, 1)
         GeneValidator.run
         $stdout.reopen original_stdout
 
@@ -77,7 +84,6 @@ module GeneValidator
 
         File.delete(prot_xml_out)
         File.delete(prot_tab_out)
-        FileUtils.rm_rf(prot_output_dir)
 
         assert_equal(true, diff)
       end
@@ -88,39 +94,45 @@ module GeneValidator
         original_stdout = $stdout.clone
         $stdout.reopen(mrna_xml_out, 'w')
 
-        FileUtils.rm_rf(mrna_output_dir) rescue Errno::ENOENT
-
         opts = {
-          validations: %w(lenc lenr frame merge dup orf align),
+          validations: %w[lenc lenr frame merge dup orf align],
           db: database,
           num_threads: threads,
-          fast: false,
+          mafft_threads: 1,
           input_fasta_file: mrna_input,
           blast_xml_file: mrna_xml,
           raw_sequences: mrna_raw,
-          test: true
+          output_dir: mrna_output_dir,
+          output_formats: %w[html csv json stdout],
+          min_blast_hits: 5,
+          test: true,
+          hide_summary: true,
+          force_rewrite: true
         }
 
-        GeneValidator.init(opts, 1, false)
+        GeneValidator.init(opts, 1)
         GeneValidator.run
         $stdout.reopen original_stdout
         $stdout.reopen(mrna_tab_out, 'w')
 
-        FileUtils.rm_rf(mrna_output_dir) rescue Errno::ENOENT
-
         opts1 = {
-          validations: %w(lenc lenr frame merge dup orf align),
+          validations: %w[lenc lenr frame merge dup orf align],
           db: database,
           num_threads: threads,
-          fast: false,
+          mafft_threads: 1,
           input_fasta_file: mrna_input,
           blast_tabular_file: mrna_tab,
           blast_tabular_options: tab_options,
           raw_sequences: mrna_raw,
-          test: true
+          output_dir: mrna_output_dir,
+          output_formats: %w[html csv json stdout],
+          min_blast_hits: 5,
+          test: true,
+          hide_summary: true,
+          force_rewrite: true
         }
 
-        GeneValidator.init(opts1, 1, false)
+        GeneValidator.init(opts1, 1)
         GeneValidator.run
         $stdout.reopen original_stdout