genevalidator 1.6.1 → 1.6.2

data/lib/genevalidator/clusterization.rb

@@ -13,7 +13,7 @@ module GeneValidator
     end
 
     def print
-      puts "Cluster: #{x} #{y}"
+      $stderr.puts "Cluster: #{x} #{y}"
     end
 
     ##
@@ -66,7 +66,7 @@ module GeneValidator
 
     def print
       objects.each do |elem|
-        puts "(#{elem[0].x},#{elem[0].y}): #{elem[1]}"
+        $stderr.puts "(#{elem[0].x},#{elem[0].y}): #{elem[1]}"
       end
     end
 
@@ -270,11 +270,11 @@ module GeneValidator
     ##
     # Prints the current cluster
     def print
-      puts "Cluster: mean = #{mean}, density = #{density}"
+      $stderr.puts "Cluster: mean = #{mean}, density = #{density}"
       lengths.sort { |a, b| a <=> b }.each do |elem|
-        puts "#{elem[0]}, #{elem[1]}"
+        $stderr.puts "#{elem[0]}, #{elem[1]}"
       end
-      puts '--------------------------'
+      $stderr.puts '--------------------------'
     end
 
     ##
@@ -334,7 +334,7 @@ module GeneValidator
       # initially each length belongs to a different cluster
       histogram.each do |elem|
         if debug
-          puts "pair (#{elem[0].x} #{elem[0].y}) appears #{elem[1]} times"
+          $stderr.puts "pair (#{elem[0].x} #{elem[0].y}) appears #{elem[1]} times"
         end
         hash = { elem[0] => elem[1] }
         cluster = PairCluster.new(hash)
@@ -353,7 +353,7 @@ module GeneValidator
         break if no_clusters != 0 && clusters.length == no_clusters
 
         iteration = iteration + 1
-        puts "\nIteration #{iteration}" if debug
+        $stderr.puts "\nIteration #{iteration}" if debug
 
         min_distance = 100_000_000
         cluster1     = 0
@@ -363,7 +363,7 @@ module GeneValidator
         [*(0..(clusters.length - 2))].each do |i|
           [*((i + 1)..(clusters.length - 1))].each do |j|
             dist = clusters[i].distance(clusters[j], distance_method)
-            puts "distance between clusters #{i} and #{j} is #{dist}" if debug
+            $stderr.puts "distance between clusters #{i} and #{j} is #{dist}" if debug
             current_density = clusters[i].density + clusters[j].density
             if dist < min_distance
               min_distance = dist
@@ -379,14 +379,14 @@ module GeneValidator
         end
 
         # merge clusters 'cluster1' and 'cluster2'
-        puts "clusters to merge #{cluster1} and #{cluster2}" if debug
+        $stderr.puts "clusters to merge #{cluster1} and #{cluster2}" if debug
 
         clusters[cluster1].add(clusters[cluster2])
         clusters.delete_at(cluster2)
 
         if debug
           clusters.each_with_index do |elem, i|
-            puts "cluster #{i}"
+            $stderr.puts "cluster #{i}"
             elem.print
           end
         end
@@ -433,7 +433,7 @@ module GeneValidator
       # clusters = array of clusters
       # initially each length belongs to a different cluster
       histogram.sort { |a, b| a[0] <=> b[0] }.each do |elem|
-        puts "len #{elem[0]} appears #{elem[1]} times" if debug
+        $stderr.puts "len #{elem[0]} appears #{elem[1]} times" if debug
         hash = { elem[0] => elem[1] }
         cluster = Cluster.new(hash)
         clusters.push(cluster)
@@ -452,7 +452,7 @@ module GeneValidator
         break if no_clusters != 0 && clusters.length == no_clusters
 
         iteration = iteration + 1
-        puts "\nIteration #{iteration}" if debug
+        $stderr.puts "\nIteration #{iteration}" if debug
 
         min_distance = 100_000_000
         cluster      = 0
@@ -460,7 +460,7 @@ module GeneValidator
 
         clusters[0..clusters.length - 2].each_with_index do |_item, i|
           dist = clusters[i].distance(clusters[i + 1], distance_method)
-          puts "distance between clusters #{i} and #{i + 1} is #{dist}" if debug
+          $stderr.puts "distance between clusters #{i} and #{i + 1} is #{dist}" if debug
           current_density = clusters[i].density + clusters[i + 1].density
           if dist < min_distance
             min_distance = dist
@@ -479,14 +479,14 @@ module GeneValidator
         end
 
         # merge clusters 'cluster' and 'cluster'+1
-        puts "clusters to merge #{cluster} and #{cluster + 1}" if debug
+        $stderr.puts "clusters to merge #{cluster} and #{cluster + 1}" if debug
 
         clusters[cluster].add(clusters[cluster + 1])
         clusters.delete_at(cluster + 1)
 
         if debug
           clusters.each_with_index do |elem, i|
-            puts "cluster #{i}"
+            $stderr.puts "cluster #{i}"
             elem.print
           end
         end

data/lib/genevalidator/exceptions.rb

@@ -3,9 +3,14 @@ module GeneValidator
   class ClasspathError < Exception
   end
 
+
   # Exception raised when the command line type argument
   # does not corrsepond to the type of the sequences in the fasta file
   class SequenceTypeError < Exception
+    def to_s
+      "\nSequence Type error: Possible cause include that the blast output" \
+      " was not obtained against a protein database.\n"
+    end
   end
 
   # Exception raised when an unexisting file is accessed
@@ -18,17 +23,29 @@ module GeneValidator
 
   # Exception raised when a validation class is not instance of ValidationTest
   class ValidationClassError < Exception
+    def to_s
+      "\nClass Type error: Possible cause include that one of the validations" \
+      " is not a sub-class of ValidationTest\n"
+    end
   end
 
-  # Exception raised when a validation report class is not instance of ValidationReport
+  # Exception raised when a validation report class is not instance of
+  #   ValidationReport
   class ReportClassError < Exception
+    def to_s
+      "\nClass Type error: Possible causes include that the type of one of" \
+      ' the validation reports is not a subclass of the ValidationReport' \
+      " class.\n"
+    end
   end
 
-  # Exception raised when there are not enough blast hits to make a statisticl validation
+  # Exception raised when there are not enough blast hits to make a statistical
+  #   validation
   class NotEnoughHitsError < Exception
   end
 
-  # Exception raised when function dependig on the internet connection raise Exception
+  # Exception raised when function dependig on the internet connection raise
+  #   Exception
   class NoInternetError < Exception
   end
 
@@ -38,21 +55,31 @@ module GeneValidator
 
   # Exception raised when the -v argument didn't filter any validatio test
   class NoValidationError < Exception
+    def to_s
+      "\nValidation error: Possible cause inlcude that the -v arguments" \
+      " supplied is not valid\n"
+    end
   end
 
   # Exception raised when the are alias duplications
   class AliasDuplicationError < Exception
+    def to_s
+      "\nAlias Duplication error: Possible cause: At least two validations" \
+      " have the same CLI alias\n"
+    end
   end
 
   # Exception raised when the are alias duplications
   class NoPIdentError < Exception
   end
 
-  # Exception raised when the tabular format does not correspond to the tabular argumet
+  # Exception raised when the tabular format does not correspond to the tabular
+  #   argument
   class InconsistentTabularFormat < Exception
   end
 
-  # Exception raised when there are more than one reading frame among the hits of one prediction
+  # Exception raised when there are more than one reading frame among the hits
+  #   of one prediction
   class ReadingFrameError < Exception
   end
 

data/lib/genevalidator/get_raw_sequences.rb

@@ -1,36 +1,73 @@
-require 'genevalidator/sequences'
-require 'genevalidator/exceptions'
 require 'bio-blastxmlparser'
+require 'forwardable'
 require 'net/http'
-require 'open-uri'
 require 'uri'
-require 'io/console'
 require 'yaml'
+
+require 'genevalidator/exceptions'
+require 'genevalidator/sequences'
+
 module GeneValidator
   # Gets the raw sequences for each hit in a BLAST output file
-  module GetRawSequences
+  class RawSequences
     class <<self
+      extend Forwardable
+      def_delegators GeneValidator, :opt, :config
+
       ##
       # Obtains raw_sequences from BLAST output file...
-      def run(opt)
+      def run
         @opt = opt
+        @config = config
 
-        if opt[:blast_xml_file]
-          @blast_file  = opt[:blast_xml_file]
+        $stderr.puts 'Extracting sequences within the BLAST output file from' \
+                     ' the BLAST database'
+
+        if @opt[:blast_xml_file]
+          @blast_file  = @opt[:blast_xml_file]
         else
-          @blast_file = opt[:blast_tabular_file]
+          @blast_file = @opt[:blast_tabular_file]
         end
 
-        raw_seq_file = @blast_file + '.raw_seq'
-        index_file   = @blast_file + '.index'
+        @opt[:raw_sequences] = @blast_file + '.raw_seq'
+        index_file           = @blast_file + '.index'
 
         if opt[:db] =~ /remote/
-          write_an_raw_seq_file(raw_seq_file, 'remote')
+          write_a_raw_seq_file(@opt[:raw_sequences], 'remote')
         else
           write_an_index_file(index_file, 'local')
-          obtain_raw_seqs_from_local_db(index_file, raw_seq_file)
+          obtain_raw_seqs_from_local_db(index_file, @opt[:raw_sequences])
+        end
+        index_raw_seq_file(@opt[:raw_sequences])
+      end
+
+      ##
+      # Index the raw sequences file...
+      def index_raw_seq_file(raw_seq_file = opt[:raw_sequences])
+        # leave only the identifiers in the fasta description
+        content = File.open(raw_seq_file, 'rb').read.gsub(/ .*/, '')
+        File.open(raw_seq_file, 'w+') { |f| f.write(content) }
+
+        # index the fasta file
+        keys   = content.scan(/>(.*)\n/).flatten
+        values = content.enum_for(:scan, /(>[^>]+)/).map { Regexp.last_match.begin(0) }
+
+        # make an index hash
+        index_hash = {}
+        keys.each_with_index do |k, i|
+          start = values[i]
+          endf  = (i == values.length - 1) ? content.length - 1 : values[i + 1]
+          index_hash[k] = [start, endf]
+        end
+
+        # create FASTA index
+        config[:raw_seq_file_index] = "#{raw_seq_file}.idx"
+        config[:raw_seq_file_load]  = index_hash
+
+        File.open(config[:raw_seq_file_index], 'w') do |f|
+          YAML.dump(index_hash, f)
         end
-        raw_seq_file
+        content = nil
       end
 
       private
@@ -43,13 +80,13 @@ module GeneValidator
         file.close unless file.nil?
       end
 
-      alias_method :write_an_raw_seq_file, :write_an_index_file
+      alias_method :write_a_raw_seq_file, :write_an_index_file
 
       def iterate_xml(file, db_type)
         n = Bio::BlastXMLParser::XmlIterator.new(@opt[:blast_xml_file]).to_enum
         n.each do |iter|
           iter.each do |hit|
-            if db_type == 'remote'
+            if db_type == 'remote' || hit.hit_id.nil?
               file.puts obtain_raw_seqs_from_remote_db(hit.accession)
             else
               file.puts hit.hit_id
@@ -57,10 +94,10 @@ module GeneValidator
           end
         end
       rescue
-        puts '*** Error: There was an error in analysing the BLAST XML file.'
-        puts '    Please ensure that BLAST XML file is in the correct format'
-        puts '    and then try again. If you are using a remote database,'
-        puts '    please ensure that you have internet access.'
+        $stderr.puts '*** Error: There was an error in analysing the BLAST XML file.'
+        $stderr.puts '    Please ensure that BLAST XML file is in the correct format'
+        $stderr.puts '    and then try again. If you are using a remote database,'
+        $stderr.puts '    please ensure that you have internet access.'
         exit 1
       end
 
@@ -73,23 +110,23 @@ module GeneValidator
         assert_table_has_correct_no_of_collumns(rows, table_headers)
 
         rows.each do |row|
-          if db_type == 'remote'
+          if db_type == 'remote' || row['sseqid'].nil?
             file.puts obtain_raw_seqs_from_remote_db(row['sacc'])
           else
             file.puts row['sseqid']
           end
         end
       rescue
-        puts '*** Error: There was an error in analysing the BLAST tabular'
-        puts '    file. Please ensure that BLAST tabular file is in the correct'
-        puts '    format and then try again. If you are using a remote'
-        puts '    database, please ensure that you have internet access.'
+        $stderr.puts '*** Error: There was an error in analysing the BLAST tabular'
+        $stderr.puts '    file. Please ensure that BLAST tabular file is in the correct'
+        $stderr.puts '    format and then try again. If you are using a remote'
+        $stderr.puts '    database, please ensure that you have internet access.'
         exit 1
       end
 
       def obtain_raw_seqs_from_local_db(index_file, raw_seq_file)
-        cmd = "blastdbcmd -entry_batch #{index_file} -db #{@opt[:db]} -outfmt" \
-              " '%f' -out #{raw_seq_file}"
+        cmd = "blastdbcmd -entry_batch '#{index_file}' -db '#{@opt[:db]}'" \
+              " -outfmt '%f' -out '#{raw_seq_file}'"
         `#{cmd}`
       end
 
@@ -106,8 +143,8 @@ module GeneValidator
         result   = Net::HTTP.get(URI.parse(uri))
         raw_seqs = result[0..result.length - 2]
         unless raw_seqs.downcase.index(/error/).nil?
-          puts '*** Error: There was an error in obtaining the raw sequence' \
-               ' of a BLAST hit. Please ensure that you have internet access.'
+          $stderr.puts '*** Error: There was an error in obtaining the raw sequence' \
+                       ' of a BLAST hit. Please ensure that you have internet access.'
           exit 1
         end
         raw_seqs
@@ -116,10 +153,10 @@ module GeneValidator
       def assert_table_has_correct_no_of_collumns(rows, table_headers)
         rows.each do |row|
           unless row.length == table_headers.length
-            puts '*** Error: The BLAST tabular file cannot be parsed. This is' \
-                 ' could possibly be due to an incorrect BLAST tabular' \
-                 ' options ("-o", "--blast_tabular_options") being supplied.' \
-                 ' Please correct this and try again.'
+            $stderr.puts '*** Error: The BLAST tabular file cannot be parsed. This is' \
+                         ' could possibly be due to an incorrect BLAST tabular' \
+                         ' options ("-o", "--blast_tabular_options") being supplied.' \
+                         ' Please correct this and try again.'
             exit 1
           end
           break # break after checking the first column

data/lib/genevalidator/hsp.rb

@@ -33,8 +33,7 @@ module GeneValidator
     # Params:
     # +column+: String with column name.
     # +value+: Value of the column
-    # +type+: type of the sequences: :nucleotide or :protein
-    def init_tabular_attribute(hash, _type = :protein)
+    def init_tabular_attribute(hash)
       @match_query_from    = hash['qstart'].to_i if hash['qstart']
       @match_query_to      = hash['qend'].to_i if hash['qend']
       @query_reading_frame = hash['qframe'].to_i if hash['qframe']
@@ -47,12 +46,10 @@ module GeneValidator
       @identity            = hash['nident'].to_f if hash['nident']
       @hsp_evalue          = hash['evalue'].to_f if hash['evalue']
       if hash['qseq']
-        puts @query_alignment
         query_seq_type = BlastUtils.guess_sequence_type(@query_alignment)
         fail SequenceTypeError if query_seq_type != :protein
       end
       if hash['sseq']
-        puts @hit_alignment
         hit_seq_type = BlastUtils.guess_sequence_type(@hit_alignment)
         fail SequenceTypeError if hit_seq_type != :protein
       end

data/lib/genevalidator/json_to_gv_results.rb

@@ -0,0 +1,109 @@
+require 'erb'
+require 'fileutils'
+require 'forwardable'
+require 'json'
+
+require 'genevalidator'
+require 'genevalidator/output'
+require 'genevalidator/version'
+
+module GeneValidator
+  # produce GV results from a JSON previously produced from GV
+  class JsonToGVResults
+    class << self
+      extend Forwardable
+      def_delegators GeneValidator, :opt
+
+      def init
+        @opt    = opt
+        @config = {
+          html_path: "#{@opt[:json_file]}.html",
+          plot_dir: "#{@opt[:json_file]}.html/files/json",
+          aux: File.expand_path(File.join(File.dirname(__FILE__), '../../aux')),
+          filename: File.basename(@opt[:json_file]),
+          output_max: 2500,
+          run_no: 0
+        }
+        @json_array = load_json_file
+      end
+
+      def run
+        init
+        GeneValidator.create_output_folder(@config[:html_path], @config[:aux])
+        @json_array.each do |row|
+          @config[:run_no] += 1
+          create_json_file(row)
+          output_html = output_filename
+          generate_html_header(output_html) unless File.exist?(output_html)
+          generate_html_query(output_html, row)
+        end
+        html_footer
+      end
+
+      def load_json_file
+        json_contents = File.read(File.expand_path(@opt[:json_file]))
+        JSON.load(json_contents)
+      end
+
+      def create_json_file(row)
+        @json_file = File.join(@config[:plot_dir],
+                               "#{@config[:filename]}_#{row['idx']}.json")
+        File.open(@json_file, 'w') { |f| f.write(row.to_json) }
+      end
+
+      def output_filename
+        i = (@config[:run_no].to_f / @config[:output_max]).ceil
+        File.join(@config[:html_path], "results#{i}.html")
+      end
+
+      def generate_html_header(output_html)
+        return if File.exist?(output_html)
+        json_header_template = File.join(@config[:aux], 'json_header.erb')
+        template_contents    = File.open(json_header_template, 'r').read
+        erb                  = ERB.new(template_contents, 0, '>')
+        File.open(output_html, 'w+') { |f| f.write(erb.result(binding)) }
+      end
+
+      def generate_html_query(output_html, row)
+        @row = row
+        json_query_template = File.join(@config[:aux], 'json_query.erb')
+        template_contents   = File.open(json_query_template, 'r').read
+        erb                 = ERB.new(template_contents, 0, '>')
+        File.open(output_html, 'a') { |f| f.write(erb.result(binding)) }
+      end
+
+      # Add footer to all output files
+      def html_footer
+        no_of_output_files = (@config[:run_no].to_f / @config[:output_max]).ceil
+
+        output_files = []
+        (1..no_of_output_files).each { |i| output_files << "results#{i}.html" }
+
+        write_html_footer(no_of_output_files, output_files)
+      end
+
+      def write_html_footer(no_of_output_files, output_files)
+        turn_off_automated_sorting
+        json_footer_template = File.join(@config[:aux], 'json_footer.erb')
+        template_contents    = File.open(json_footer_template, 'r').read
+        erb                  = ERB.new(template_contents, 0, '>')
+        (1..no_of_output_files).each do |i|
+          results_html = File.join(@config[:html_path], "results#{i}.html")
+          File.open(results_html, 'a+') { |f| f.write(erb.result(binding)) }
+        end
+      end
+
+      # Since the whole idea is that users would sort by
+      def turn_off_automated_sorting
+        script_file = File.join(@config[:html_path], 'files/js/script.js')
+        temp_file   = File.join(@config[:html_path], 'files/js/script.temp.js')
+        File.open(temp_file, 'w') do |out_file|
+          out_file.puts File.readlines(script_file)[0..23].join
+          out_file.puts '}'
+          out_file.puts File.readlines(script_file)[26..-1].join
+        end
+        FileUtils.mv(temp_file, script_file)
+      end
+    end
+  end
+end