full_lengther_next 0.6.2 → 0.9.8

data/History.txt

@@ -1,32 +0,0 @@
-=== 0.1.0 2013-09-12
-
-Major rewrite of script
-
-=== 0.0.7 2012-07-25
-
-Chimera detection
-
-=== 0.0.6 2012-04-16
-
-Fixed some cosmetic issues and parameters names
-
-=== 0.0.5 2012-03-09
-
-Fix NCRNA annotation
-
-=== 0.0.4 2012-03-07
-
-Fixed stats for 0 seqs
-
-=== 0.0.3 2012-03-01
-
-Added ncrna
-
-=== 0.0.2 2012-02-07
-
-Added FULL_LENGTH_NEXT_INIT environment variable for clustered installations
-
-=== 0.0.1 2012-01-30
-
-* 1 major enhancement:
-  * Initial release

data/Manifest.txt

@@ -1,44 +0,0 @@
-Rakefile
-script
-script/generate
-script/destroy
-script/console
-test
-test/test_full_lengther_next.rb
-test/test_helper.rb
-bin/plot_taxonomy.rb
-bin/plot_fln.rb
-bin/download_fln_dbs.rb
-bin/full_lengther_next
-bin/make_user_db.rb
-bin/make_test_dataset.rb
-PostInstall.txt
-README.rdoc
-History.txt
-Manifest.txt
-lib/full_lengther_next
-lib/full_lengther_next/classes
-lib/full_lengther_next/classes/blast_functions.rb
-lib/full_lengther_next/classes/my_worker_manager_fln.rb
-lib/full_lengther_next/classes/types.rb
-lib/full_lengther_next/classes/chimeric_seqs.rb
-lib/full_lengther_next/classes/artifacts.rb
-lib/full_lengther_next/classes/cdhit.rb
-lib/full_lengther_next/classes/fl_analysis.rb
-lib/full_lengther_next/classes/fl_string_utils.rb
-lib/full_lengther_next/classes/my_worker.rb
-lib/full_lengther_next/classes/sequence.rb
-lib/full_lengther_next/classes/my_worker_EST.rb
-lib/full_lengther_next/classes/test_code.rb
-lib/full_lengther_next/classes/orf.rb
-lib/full_lengther_next/classes/une_los_hit.rb
-lib/full_lengther_next/classes/warnings.rb
-lib/full_lengther_next/classes/fln_stats.rb
-lib/full_lengther_next/classes/my_worker_manager_EST.rb
-lib/full_lengther_next/classes/nc_rna.rb
-lib/full_lengther_next/classes/reptrans.rb
-lib/full_lengther_next/classes/common_functions.rb
-lib/full_lengther_next/classes/exonerate_result.rb
-lib/full_lengther_next/classes/handle_db.rb
-lib/full_lengther_next.rb
-lib/expresscanvas.zip

data/PostInstall.txt

@@ -1,6 +0,0 @@
-For more information on full_lengther_next, see * http://www.scbi.uma.es/downloads
-
-Now it is time to install databases using the following command:
-
-$> algo
-

data/bin/plot_fln.rb

@@ -1,270 +0,0 @@
-#!/usr/bin/env ruby
-require 'optparse'
-
-#############################################
-### FUNCTIONS
-#############################################
-def create_fln_hash(path)
-  fln_hash = {}
-  file = File.open(path, 'r').each do |line|
-    fields = line.chomp.split
-    fln_hash[fields[1]] = fields[0].to_i
-  end
-  fln_hash['<=200seqs'] = fln_hash['good_seqs'] - fln_hash['sequences_>200']  
-  fln_hash['>200seqs'] = fln_hash['sequences_>200']  - fln_hash['sequences_>500']
-  fln_hash['<=200unk'] = fln_hash['unknown'] - fln_hash['unknown_>200']  
-  fln_hash['>200unk'] = fln_hash['unknown_>200']  - fln_hash['unknown_>500']
-  fln_hash['<=200cod'] = fln_hash['coding'] - fln_hash['coding_>200']  
-  fln_hash['>200cod'] = fln_hash['coding_>200']  - fln_hash['coding_>500']
-  fln_hash['no_match_db'] = fln_hash['coding'] + fln_hash['unknown']
-
-  return fln_hash
-end
-
-def graph_table(fln_hash, output, graph_type, header_titles, categories_names, keywords, stacked_cols, titles = nil)
-  table = []
-  cmd = basic_plot_command(graph_type)
-  cmd << "set output '#{output}.png'\n"
-
-  if fln_hash.class.to_s == 'Array'
-    table << header(fln_hash.length, header_titles)
-    table.concat(categories(categories_names))
-    cmd << 'plot '
-    count = 0
-    fln_hash.each_with_index do |hash,i|
-      table = fill_table(hash, table, keywords, stacked_cols, graph_type)
-      if i == 0
-        first = TRUE
-      else
-        first = FALSE
-      end
-      cmd << histogram(stacked_cols, output,titles[i], count, graph_type, first)
-      if i < fln_hash.length-1
-        cmd << "\\\n"
-      end
-      if !graph_type.include?('clustered')
-        count += stacked_cols
-      else
-        count += 1
-      end
-    end
-  else
-    table << header(1, header_titles)
-    table.concat(categories(categories_names))
-    table = fill_table(fln_hash, table, keywords, stacked_cols, graph_type)
-    cmd << 'plot '+ histogram(stacked_cols, output, '', 0, graph_type, TRUE)
-  end
-  if table.length ==2 #Dummie row for rowstacked graph with a only category
-    table << table[1].dup
-    table[2].each_with_index do |cell, i|
-        if i== 0
-          table[2][i] = '&'
-        else
-          table[2][i] = 0
-        end
-    end
-  end
-  cmd.chop!
-  write_table(table, output)
-  write_cmd(cmd)
-  system('gnuplot cmd.dem')
-end
-
-def histogram(columns, file, name, add, graph_type, first)
-  cmd = ""
-  if first
-    cmd << "newhistogram \"#{name}\", '#{file}' using 2:xtic(1) t col,"
-  else
-    cmd << "newhistogram \"#{name}\", '' using #{2 + add}:xtic(1)  t col,"
-  end
-  if !graph_type.include?('clustered')
-    (columns-1).times do |col|
-        cmd << " '' u #{3+col+add} t col,"
-    end
-  end
-  return cmd
-end
-
-def write_cmd(cmd)
-  if File.exists?('cmd.dem')
-    File.delete('cmd.dem')
-  end
-  file = File.open('cmd.dem', 'a')
-  file.puts cmd
-  file.close
-end
-
-def header(iterations, header_titles)
-  header = ['Clasification']
-  iterations.times do
-    header_titles.each do |title|
-      header << title
-    end
-  end
-  return header
-end
-
-def categories(cat)
-  array_cat = cat.split(' ').map{|name| [name]}
-  return array_cat
-end
-
-def fill_table(fln_hash, table, keywords, stacked_cols,graph_type)
-  series = 0
-  keywords.each_with_index do |key, i|
-    if graph_type.include?('clustered')
-      row = i +1 -stacked_cols*series
-      if (i+1) % stacked_cols == 0
-        series +=1
-      end
-    else
-      row = i/stacked_cols + 1
-    end
-    value = fln_hash[key]
-    if value.nil?
-      value = 0
-    end
-    if table.length == 2
-      table[1] << value
-    else
-      table[row] << value 
-    end
-  end
-  return table
-end
-
-def write_table(table, file_name)
-  file_table = File.open(file_name, 'w')
-    table.each do |line|
-      file_table.puts line.join(' ')
-    end
-  file_table.close
-end
-
-def basic_plot_command(graph_type)
-  cmd = ''
-  if graph_type.include?('clustered')
-    cmd << "unset key\n"
-  else
-    cmd << "set key under nobox\n"
-  end
-  cmd << "set style data histogram\n"
-  cmd << "set style histogram #{graph_type} title offset 2,0.25\n"
-  cmd << "set style fill solid noborder\n"
-  cmd << "set boxwidth 0.95\n"
-  cmd << "unset xtics\n"
-  cmd << "set xtics nomirror rotate by -45 scale 0\n"
-  cmd << "set xlabel \" \" offset 0,-2\n"
-  cmd << "set ylabel \"Num sequences\"\n"
-  cmd << "set ytics\n"
-  cmd << "set grid y\n"
-  cmd << "set auto y\n"
-  cmd << "set terminal png nocrop enhanced font arial 15 size 1000,600\n" 
-  return cmd
-end
-
-
-
-def parse_file(file)
-  titles = []
-  paths =[]
-  File.open(file,'r').each do |line|
-    fields = line.chomp.split("\t")
-    if  !fields[0].nil?
-      titles << fields[0]
-    end
-    if !fields[1].nil?
-      paths << fields[1]
-    end
-  end
-
-  return titles, paths
-end
-
-
-##########################################################################################
-## OPTIONS
-##########################################################################################
-
-options = {}
-
-optparse = OptionParser.new do |opts|
-  options[:file]='samples'
-  opts.on( '-f', '--file FILE', 'Path to FLN execution') do |file|
-          options[:file]=file
-  end
-
-  options[:path] = File.join('fln_results','summary_stats.txt')
-  opts.on( '-p', '--path PATH', 'Path to FLN different FLN results' ) do |path|
-          options[:path] = File.join(path,'fln_results','summary_stats.txt')
-  end
-
-  # Set a banner, displayed at the top of the help screen.
-  opts.banner = "Usage: plot_fln.rb [-p PATH || -f FILE]  \n\n"
-
-  # This displays the help screen
-  opts.on( '-h', '--help', 'Display this screen' ) do
-    puts opts
-    exit
-  end
-
-end # End opts
-
-# parse options and remove from ARGV
-optparse.parse!
-
-##########################################################################################
-## MAIN
-##########################################################################################
-
-if File.exists?(options[:path]) 
-  fln_hash = create_fln_hash(options[:path])
-end
-
-if File.exists?(options[:file])
-  titles, paths = parse_file(options[:file])
-  fln_hash = []
-  paths.each do |path| 
-    fln_hash << create_fln_hash(File.join(path,'fln_results','summary_stats.txt'))
-  end
-end
-
-graph_table(
-  fln_hash, 
-  'status_report_table',
-  'rowstacked', 
-  %w{Sure Putative}, 
-  'Complete N-terminal C-terminal Internal NcRNA Coding Unknown', 
-  %w{complete_sure complete_putative n_terminal_sure n_terminal_putative c_terminal_sure c_terminal_putative internal internal_putative ncrna ncrna_putative coding_sure coding_putative unknown unknown_putative}, 
-  2, 
-  titles)
-
-graph_table(
-  fln_hash,
-  'assembly_table',
-  'rowstacked', 
-  %w{<=200nt >200nt >500nt}, 
-  'Unigenes Coding Unknown', 
-  %w{<=200seqs >200seqs sequences_>500 <=200cod >200cod coding_>500 <=200unk >200unk unknown_>500}, 
-  3, 
-  titles)
-
-graph_table(
-  fln_hash,
-  'database_usage',
-  'clustered', 
-  %w{seqs}, 
-  'UserDB SwissProt TrEMBL ncRNA None Diff-orthologues Complete Diff-complete', 
-  %w{userdb swissprot trembl ncrna no_match_db different_orthologues complete different_completes}, 
-  8, 
-  titles)
-
-graph_table(
-  fln_hash,
-  'artifacts',
-  'clustered', 
-  %w{seqs}, 
-  'Misassembled Chimeras Other', 
-  %w{misassembled chimeras other_artifacts}, 
-  3, 
-  titles)

data/bin/plot_taxonomy.rb

@@ -1,70 +0,0 @@
-#!/usr/bin/env ruby
-
-require 'scbi_plot'
-#############################################
-### FUNCTIONS
-#############################################
-def take_taxonomy(file)
-	taxonomy = {}
-	File.open(file).each_with_index do |line, i|
-		line.chomp!
-		field = line.split("\t").last
-		organism = field.split(";",2).last
-		organism.gsub!(/\(\D+\)/,'')
-		if organism.split(' ').length == 1
-			next
-		end
-		organism.gsub!('.','')
-		organism.gsub!(/^ /,'')
-		organism.gsub!('  ','')
-		organism ='"'+organism+'"'
-		if taxonomy[organism].nil?
-			taxonomy[organism] = 1
-		else
-			taxonomy[organism] += 1
-		end	
-	end
-	return taxonomy
-end
-
-def plot(taxonomy)
-	p=ScbiPlot::Histogram.new('fln_taxonomy_plot.png','Group organism representation')
-	p.add_x(taxonomy.keys)
-	p.add_y(taxonomy.values)
-	p.do_graph
-end
-
-#############################################
-
-require 'optparse'
-options = {}
-
-optparse = OptionParser.new do |opts|
-   options[:path] = File.join('fln_results','pt_seqs')
-  opts.on( '-p', '--path PATH', 'Path to FLN execution' ) do |path|
-          options[:path] = File.join(path,'fln_results','pt_seqs')
-  end
-
-  # Set a banner, displayed at the top of the help screen.
-  opts.banner = "Usage: plot_taxonomy.rb -p PATH  \n\n"
-
-  # This displays the help screen
-  opts.on( '-h', '--help', 'Display this screen' ) do
-    puts opts
-    exit
-  end
-
-end # End opts
-
-# parse options and remove from ARGV
-optparse.parse!
-
-taxonomy = nil
-if File.exists?(options[:path])
-	taxonomy = take_taxonomy(options[:path])
-else
-	puts 'Path isn\'t valid'
-	Process.exit
-end
-
-plot(taxonomy)

data/lib/full_lengther_next/classes/artifacts.rb

@@ -1,66 +0,0 @@
-require 'blast_functions'
-require 'types'
-require 'chimeric_seqs'
-include ChimericSeqs
-
-#####################################################################
-## MAIN FUNCTION
-#####################################################################
-def artifact?(seq, query, db_name, db_path, options, new_seqs)
-	artifact = FALSE
-	# MISASSEMBLED DETECTION
-	if misassembled_detection(query) #If seq is misassembled stop chimera analisys
-		seq.hit = query.hits.first
-		artifact = TRUE
-		seq.type = MISASSEMBLED
-		seq.warnings('ERROR#1')
-	end
-
-	# OVERLAPPING HSPS ON SUBJECT DETECTION
-=begin
-	if !artifact
-		hit_reference = query.hits.first.dup
-		query, overlapping = overlapping_hsps_on_subject(query)
-		if overlapping
-			if query.hits.first.nil?
-				seq.hit = hit_reference
-			else
-				seq.hit = query.hits.first
-			end
-			artifact = TRUE
-			seq.type = OTHER
-			seq.warnings('ERROR#2')
-		end
-	end
-=end
-
-	# MULTIPLE HSP DETECTION
-	if !artifact && multiple_hsps(query, 3)   
-		seq.hit = query.hits.first
-		seq.warnings('ERROR#3')
-	end
-
-	# CHIMERA DETECTION
-	if !artifact && !options[:chimera].include?('d')  
-		chimera = search_chimeras(seq, query, options, db_name, db_path)			
-		if !chimera.nil?   
-			new_seqs.concat(chimera)
-			seq.db_name = db_name
-			seq.type = CHIMERA
-			artifact = TRUE
-		end
-	end
-
-	if artifact
-		if $verbose > 1
-			puts seq.prot_annot_calification
-		end
-		seq.db_name = db_name
-		seq.save_fasta = FALSE
-		seq.ignore = TRUE
-	end
-	return artifact
-end
-
-
-