macroape 3.3.3 → 3.3.4

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: macroape
 version: !ruby/object:Gem::Version
-  version: 3.3.3
+  version: 3.3.4
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-07-19 00:00:00.000000000 Z
+date: 2012-09-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bioinform
@@ -18,7 +18,7 @@ dependencies:
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
-        version: 0.1.2
+        version: 0.1.5
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
@@ -26,7 +26,7 @@ dependencies:
     requirements:
     - - ! '>='
       - !ruby/object:Gem::Version
-        version: 0.1.2
+        version: 0.1.5
 description: Macroape is an abbreviation for MAtrix CompaRisOn by Approximate P-value
   Estimation. It's a bioinformatic tool for evaluating similarity measure and best
   alignment between a pair of Position Weight Matrices(PWM), finding thresholds by
@@ -35,6 +35,7 @@ description: Macroape is an abbreviation for MAtrix CompaRisOn by Approximate P-
 email:
 - prijutme4ty@gmail.com
 executables:
+- align_motifs
 - eval_alignment
 - eval_similarity
 - find_pvalue
@@ -51,6 +52,7 @@ files:
 - Rakefile.rb
 - TODO.txt
 - benchmark/similarity_benchmark.rb
+- bin/align_motifs
 - bin/eval_alignment
 - bin/eval_similarity
 - bin/find_pvalue
@@ -59,32 +61,45 @@ files:
 - bin/scan_collection
 - lib/macroape.rb
 - lib/macroape/aligned_pair_intersection.rb
+- lib/macroape/cli.rb
+- lib/macroape/cli/align_motifs.rb
+- lib/macroape/cli/eval_alignment.rb
+- lib/macroape/cli/eval_similarity.rb
+- lib/macroape/cli/find_pvalue.rb
+- lib/macroape/cli/find_threshold.rb
+- lib/macroape/cli/preprocess_collection.rb
+- lib/macroape/cli/scan_collection.rb
 - lib/macroape/collection.rb
 - lib/macroape/counting.rb
-- lib/macroape/exec/eval_alignment.rb
-- lib/macroape/exec/eval_similarity.rb
-- lib/macroape/exec/find_pvalue.rb
-- lib/macroape/exec/find_threshold.rb
-- lib/macroape/exec/preprocess_collection.rb
-- lib/macroape/exec/scan_collection.rb
 - lib/macroape/pwm_compare.rb
 - lib/macroape/pwm_compare_aligned.rb
 - lib/macroape/version.rb
 - macroape.gemspec
 - spec/count_distribution_spec.rb
 - spec/spec_helper.rb
+- test/align_motifs_test.rb
 - test/data/AHR_si.pat
+- test/data/KLF3_f1.pat
+- test/data/KLF3_f1.pcm
 - test/data/KLF4_f2.pat
+- test/data/KLF4_f2.pcm
 - test/data/KLF4_f2_scan_results_all.txt
 - test/data/KLF4_f2_scan_results_default_cutoff.txt
 - test/data/KLF4_f2_scan_results_precise_mode.txt
 - test/data/SP1_f1.pat
+- test/data/SP1_f1.pcm
 - test/data/SP1_f1_revcomp.pat
+- test/data/SP1_f1_revcomp.pcm
 - test/data/test_collection.yaml
 - test/data/test_collection/GABPA_f1.pat
 - test/data/test_collection/KLF4_f2.pat
 - test/data/test_collection/SP1_f1.pat
-- test/eval_alignment_similarity_test.rb
+- test/data/test_collection_pcm/GABPA_f1.pcm
+- test/data/test_collection_pcm/KLF4_f2.pcm
+- test/data/test_collection_pcm/SP1_f1.pcm
+- test/data/test_collection_single_file.txt
+- test/data/test_collection_single_file_pcm.txt
+- test/eval_alignment_test.rb
 - test/eval_similarity_test.rb
 - test/find_pvalue_test.rb
 - test/find_threshold_test.rb
@@ -118,18 +133,29 @@ summary: PWM comparison tool using MACROAPE approach
 test_files:
 - spec/count_distribution_spec.rb
 - spec/spec_helper.rb
+- test/align_motifs_test.rb
 - test/data/AHR_si.pat
+- test/data/KLF3_f1.pat
+- test/data/KLF3_f1.pcm
 - test/data/KLF4_f2.pat
+- test/data/KLF4_f2.pcm
 - test/data/KLF4_f2_scan_results_all.txt
 - test/data/KLF4_f2_scan_results_default_cutoff.txt
 - test/data/KLF4_f2_scan_results_precise_mode.txt
 - test/data/SP1_f1.pat
+- test/data/SP1_f1.pcm
 - test/data/SP1_f1_revcomp.pat
+- test/data/SP1_f1_revcomp.pcm
 - test/data/test_collection.yaml
 - test/data/test_collection/GABPA_f1.pat
 - test/data/test_collection/KLF4_f2.pat
 - test/data/test_collection/SP1_f1.pat
-- test/eval_alignment_similarity_test.rb
+- test/data/test_collection_pcm/GABPA_f1.pcm
+- test/data/test_collection_pcm/KLF4_f2.pcm
+- test/data/test_collection_pcm/SP1_f1.pcm
+- test/data/test_collection_single_file.txt
+- test/data/test_collection_single_file_pcm.txt
+- test/eval_alignment_test.rb
 - test/eval_similarity_test.rb
 - test/find_pvalue_test.rb
 - test/find_threshold_test.rb

data/lib/macroape/exec/eval_alignment.rb

@@ -1,125 +0,0 @@
-help_string = %q{
-Command-line format:
-ruby eval_alignment.rb <1st matrix pat-file> <2nd matrix pat-file> <shift> <orientation(direct/revcomp)> [options]
-type <1st matrix pat-file> <2nd matrix pat-file> | ruby eval_alignment.rb .stdin .stdin <shift> <orientation(direct/revcomp)> [options]
-     or in linux
-cat <1st matrix pat-file> <2nd matrix pat-file> | ruby eval_alignment.rb .stdin .stdin <shift> <orientation(direct/revcomp)> [options]
-
-Options:
-  [-p <P-value>]
-  [-d <discretization level>]
-  [-b <background probabilities, ACGT - 4 numbers, space-delimited, sum should be equal to 1>]
-
-Output format:
-  <jaccard similarity coefficient>
-  <number of words recognized by both 1st and 2nd matrices | probability to draw a word recognized by both 1st and 2nd matrices> <length of the given alignment>
-  <aligned 1st matrix>
-  <aligned 2nd matrix>
-  <shift> <orientation>
-
-Examples:
-  ruby eval_alignment.rb motifs/KLF4_f2.pat motifs/SP1_f1.pat -1 direct -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
-     or on windows
-  type motifs/SP1.pat | ruby eval_alignment.rb motifs/KLF4.pat .stdin 0 revcomp -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
-     or in linux
-  cat motifs/KLF4.pat motifs/SP1.pat | ruby eval_alignment.rb .stdin .stdin 3 direct -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
-}
-$:.unshift File.join(File.dirname(__FILE__),'./../../')
-require 'macroape'
-
-if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
-  STDERR.puts help_string
-  exit
-end
-
-pvalue = 0.0005
-discretization = 10
-
-first_background = [1,1,1,1]
-second_background = [1,1,1,1]
-
-begin
-  first_file = ARGV.shift
-  second_file = ARGV.shift
-
-  shift = ARGV.shift
-  orientation = ARGV.shift
-
-  raise "You'd specify two input sources (each is filename or .stdin)" unless first_file and second_file
-  raise 'You\'d specify shift' unless shift
-  raise 'You\'d specify orientation' unless orientation
-
-  shift = shift.to_i
-  orientation = orientation.to_sym
-
-  case orientation
-    when :direct
-      reverse = false
-    when :revcomp
-      reverse = true
-    else
-      raise 'Unknown orientation(direct/revcomp)'
-  end
-
-
-  until ARGV.empty?
-    case ARGV.shift
-      when '-p'
-        pvalue = ARGV.shift.to_f
-      when '-d'
-        discretization = ARGV.shift.to_f
-      when '-m'
-        Macroape::MaxHashSizeSingle = ARGV.shift.to_f
-      when '-md'
-        Macroape::MaxHashSizeDouble = ARGV.shift.to_f
-      when '-b'
-        second_background = first_background = ARGV.shift(4).map(&:to_f)
-      when '-b1'
-        first_background = ARGV.shift(4).map(&:to_f)
-      when '-b2'
-        second_background = ARGV.shift(4).map(&:to_f)
-    end
-  end
-  raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless first_background == first_background.reverse
-  raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless second_background == second_background.reverse
-
-
-  Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
-  Macroape::MaxHashSizeDouble = 1000 unless defined? Macroape::MaxHashSizeDouble
-
-#  if first_file == '.stdin' || second_file == '.stdin'
-#    r_stream, w_stream = IO.pipe
-#    STDIN.readlines.each{|line| w_stream.write(line)}
-#    w_stream.close
-#  end
-
-  if first_file == '.stdin'
-#    r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
-#    pwm_first = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(first_background).discrete(discretization)
-  else
-    raise "Error! File #{first_file} don't exist" unless File.exist?(first_file)
-    pwm_first = Bioinform::PWM.new(File.read(first_file)).background(first_background).discrete(discretization)
-  end
-
-  if second_file == '.stdin'
-#    r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
-#    pwm_second = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(second_background).discrete(discretization)
-  else
-    raise "Error! File #{second_file} don't exist" unless File.exist?(second_file)
-    pwm_second = Bioinform::PWM.new(File.read(second_file)).background(second_background).discrete(discretization)
-  end
-
-#  r_stream.close if first_file == '.stdin' || second_file == '.stdin'
-
-  cmp = Macroape::PWMCompareAligned.new(pwm_first, pwm_second, shift, orientation)
-
-  first_threshold = pwm_first.threshold(pvalue)
-  second_threshold = pwm_second.threshold(pvalue)
-
-  info = cmp.alignment_infos.merge( cmp.jaccard(first_threshold, second_threshold) )
-
-  puts "#{info[:similarity]}\n#{info[:recognized_by_both]}\t#{info[:alignment_length]}\n#{info[:text]}\n#{info[:shift]}\t#{info[:orientation]}"
-
-rescue => err
-  STDERR.puts "\n#{err}\n#{err.backtrace.first(5).join("\n")}\n\nUse -help option for help\n"
-end

data/lib/macroape/exec/eval_similarity.rb

@@ -1,108 +0,0 @@
-help_string = %q{
-Command-line format:
-ruby eval_similarity.rb <1st matrix pat-file> <2nd matrix pat-file> [options]
-     or on windows
-type <1st matrix pat-file> <2nd matrix pat-file> | ruby eval_similarity.rb .stdin .stdin [options]
-     or in linux
-cat <1st matrix pat-file> <2nd matrix pat-file> | ruby eval_similarity.rb .stdin .stdin [options]
-
-Options:
-  [-p <P-value>]
-  [-d <discretization level>]
-  [-b <background probabilities, ACGT - 4 numbers, space-delimited, sum should be equal to 1>]
-
-Output has format:
-  <jaccard similarity coefficient>
-  <number of words recognized by both 1st and 2nd matrices | probability to draw a word recognized by both 1st and 2nd matrices> <length of the optimal alignment>
-  <optimal alignment, the 1st matrix>
-  <optimal alignment, the 2nd matrix>
-  <shift> <orientation>
-
-Examples:
-  ruby eval_similarity.rb motifs/KLF4.pat motifs/SP1.pat -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
-     or on windows
-  type motifs/SP1.pat | ruby eval_similarity.rb motifs/KLF4.pat .stdin -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
-     or in linux
-  cat motifs/KLF4.pat motifs/SP1.pat | ruby eval_similarity.rb .stdin .stdin -p 0.0005 -d 100 -b 0.4 0.3 0.2 0.1
-}
-
-$:.unshift File.join(File.dirname(__FILE__),'./../../')
-require 'macroape'
-
-if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
-  STDERR.puts help_string
-  exit
-end
-
-pvalue = 0.0005
-discretization = 10
-
-first_background = [1,1,1,1]
-second_background = [1,1,1,1]
-
-begin
-  first_file = ARGV.shift
-  second_file = ARGV.shift
-  raise "You'd specify two input sources (each is filename or .stdin)" unless first_file and second_file
-
-  until ARGV.empty?
-    case ARGV.shift
-      when '-p'
-        pvalue = ARGV.shift.to_f
-      when '-d'
-        discretization = ARGV.shift.to_f
-      when '-m'
-        Macroape::MaxHashSizeSingle = ARGV.shift.to_f
-      when '-md'
-        Macroape::MaxHashSizeDouble = ARGV.shift.to_f
-      when '-b'
-        second_background = first_background = ARGV.shift(4).map(&:to_f)
-      when '-b1'
-        first_background = ARGV.shift(4).map(&:to_f)
-      when '-b2'
-        second_background = ARGV.shift(4).map(&:to_f)
-    end
-  end
-  raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless first_background == first_background.reverse
-  raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless second_background == second_background.reverse
-
-  Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
-  Macroape::MaxHashSizeDouble = 1000 unless defined? Macroape::MaxHashSizeDouble
-
-
-#  if first_file == '.stdin' || second_file == '.stdin'
-#    r_stream, w_stream = IO.pipe
-#    STDIN.readlines.each{|line| w_stream.write(line)}
-#    w_stream.close
-#  end
-
-  if first_file == '.stdin'
-#    r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
-#    pwm_first = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(first_background).discrete(discretization)
-  else
-    raise "Error! File #{first_file} don't exist" unless File.exist?(first_file)
-    pwm_first = Bioinform::PWM.new(File.read(first_file)).background(first_background).discrete(discretization)
-  end
-
-  if second_file == '.stdin'
-#    r_stream, w_stream, extracted_pwm = extract_pwm(r_stream, w_stream)
-#    pwm_second = Macroape::SingleMatrix.load_from_line_array(extracted_pwm).with_background(second_background).discrete(discretization)
-  else
-    raise "Error! File #{second_file} don't exist" unless File.exist?(second_file)
-    pwm_second = Bioinform::PWM.new(File.read(second_file)).background(second_background).discrete(discretization)
-  end
-
-  r_stream.close if first_file == '.stdin' || second_file == '.stdin'
-
-  cmp = Macroape::PWMCompare.new(pwm_first, pwm_second)
-
-  first_threshold = pwm_first.threshold(pvalue)
-  second_threshold = pwm_second.threshold(pvalue)
-
-  info = cmp.jaccard(first_threshold, second_threshold)
-
-  puts "#{info[:similarity]}\n#{info[:recognized_by_both]}\t#{info[:alignment_length]}\n#{info[:text]}\n#{info[:shift]}\t#{info[:orientation]}"
-
-rescue => err
-  STDERR.puts "\n#{err}\n#{err.backtrace.first(5).join("\n")}\n\nUse -help option for help\n"
-end

data/lib/macroape/exec/find_pvalue.rb

@@ -1,81 +0,0 @@
-help_string = %q{
-Command-line format:
-ruby find_pvalue.rb <pat-file> <threshold list> [options]
-        	or in linux
-cat <pat-file> | ruby find_pvalue.rb .stdin <threshold> [options]
- 	or on windows
-type <pat-file> | ruby find_pvalue.rb .stdin <threshold> [options]
-
-Options:
-  [-d <discretization level>]
-  [-b <background probabilities, ACGT - 4 numbers, space-delimited, sum should be equal to 1>]
-
-Output format:
-	threshold_1 count_1  pvalue_1
-	threshold_2 count_2  pvalue_2
-	threshold_3 count_3  pvalue_3
-The results are printed out in the same order as in the given threshold list.
-
-Examples:
-  ruby find_pvalue.rb motifs/KLF4.pat 7.32 -d 1000 -b 0.2 0.3 0.2 0.3
-           or on windows
-  type motifs/KLF4.pat | ruby find_pvalue.rb .stdin 7.32 4.31 5.42
-           or in linux
-  cat motifs/KLF4.pat | ruby find_pvalue.rb .stdin 7.32 4.31 5.42
-}
-
-$:.unshift File.join(File.dirname(__FILE__),'./../../')
-require 'macroape'
-
-if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
-  STDERR.puts help_string
-  exit
-end
-
-discretization = 10000
-background = [1,1,1,1]
-thresholds = []
-begin
-  filename = ARGV.shift
-
-  loop do
-    begin
-      Float(ARGV.first)
-      thresholds << ARGV.shift.to_f
-    rescue
-      raise StopIteration
-    end
-  end
-
-  raise "No input. You'd specify input source: filename or .stdin" unless filename
-  raise 'You should specify at least one threshold' if thresholds.empty?
-
-  until ARGV.empty?
-    case ARGV.shift
-      when '-b'
-        background = ARGV.shift(4).map(&:to_f)
-      when '-d'
-        discretization = ARGV.shift.to_f
-      when '-m'
-        Macroape::MaxHashSizeSingle = ARGV.shift.to_f
-    end
-  end
-  Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
-
-
-  if filename == '.stdin'
-#  TODO
-  else
-    raise "Error! File #{filename} doesn't exist" unless File.exist?(filename)
-    pwm = Bioinform::PWM.new( File.read(filename) )
-  end
-  pwm.background(background)
-
-  counts = pwm.discrete(discretization).counts_by_thresholds(* thresholds.map{|count| count * discretization})
-  pvalues = counts.map{|count| count.to_f / pwm.vocabulary_volume}
-  pvalues.zip(thresholds,counts).each{|pvalue,threshold,count|
-    puts "#{threshold}\t#{count}\t#{pvalue}"
-  }
-rescue => err
-  STDERR.puts "\n#{err}\n#{err.backtrace.first(5).join("\n")}\n\nUse -help option for help\n"
-end

data/lib/macroape/exec/find_threshold.rb

@@ -1,77 +0,0 @@
-help_string = %q{
-Command-line format::
-ruby find_threshold.rb <pat-file> [options]
-        	or in linux
-cat <pat-file> | ruby find_threshold.rb .stdin [options]
-            or on windows
-type <pat-file> | ruby find_threshold.rb .stdin [options]
-
-Options:
-  [-p <list of P-values>]
-  [-d <discretization level>]
-  [-b <background probabilities, ACGT - 4 numbers, space-delimited, sum should be equal to 1>]
-
-Output format:
- 	requested_pvalue_1 threshold_1 achieved_pvalue_1
- 	requested_pvalue_2 threshold_2 achieved_pvalue_2
-
-
-Example:
-  ruby find_threshold.rb motifs/KLF4.pat -p 0.001 0.0001 0.0005 -d 1000 -b 0.4 0.3 0.2 0.1
-}
-
-$:.unshift File.join(File.dirname(__FILE__),'./../../')
-require 'macroape'
-
-if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
-  STDERR.puts help_string
-  exit
-end
-
-background = [1,1,1,1]
-default_pvalues = [0.0005]
-discretization = 10000
-
-begin
-  filename = ARGV.shift
-  raise "No input. You'd specify input source: filename or .stdin" unless filename
-
-  pvalues = []
-  until ARGV.empty?
-    case ARGV.shift
-      when '-b'
-        background = ARGV.shift(4).map(&:to_f)
-      when '-m'
-        Macroape::MaxHashSizeSingle = ARGV.shift.to_f
-      when '-p'
-        loop do
-          begin
-            Float(ARGV.first)
-            pvalues << ARGV.shift.to_f
-          rescue
-            raise StopIteration
-          end
-        end
-      when '-d'
-        discretization = ARGV.shift.to_f
-      end
-  end
-  pvalues = default_pvalues if pvalues.empty?
-
-  Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
-
-  if filename == '.stdin'
-##  TODO
-  else
-    raise "Error! File #{filename} doesn't exist" unless File.exist?(filename)
-    pwm = Bioinform::PWM.new( File.read(filename) )
-  end
-
-  pwm.background(background)
-
-  pwm.discrete(discretization).thresholds(*pvalues) do |pvalue, threshold, real_pvalue|
-    puts "#{pvalue}\t#{threshold / discretization}\t#{real_pvalue}"
-  end
-rescue => err
-  STDERR.puts "\n#{err}\n#{err.backtrace.first(5).join("\n")}\n\nUse -help option for help\n"
-end

data/lib/macroape/exec/preprocess_collection.rb

@@ -1,101 +0,0 @@
-help_string = %q{
-Command-line format:
-  ruby preprocess_collection.rb <folder with PWMs> [options]
-
-Options:
-  [-p <list of P-values>]
-  [-d <rough discretization> <precise discretization>]
-  [-b <background probabilities, ACGT - 4 numbers, space-delimited, sum should be equal to 1>]
-  [-o <output file>]
-  [--silent] - don't show current progress information during scan (by default this information's written into stderr)
-
-The tool stores preprocessed Macroape collection to the specified YAML-file.
-
-Example:
-  ruby preprocess_collection.rb ./motifs -p 0.001 0.0005 0.0001 -d 1 10 -b 0.2 0.3 0.2 0.3 -o collection.yaml
-}
-
-$:.unshift File.join(File.dirname(__FILE__),'./../../')
-require 'macroape'
-require 'yaml'
-
-if ARGV.empty? or ARGV.include? '-h' or ARGV.include? '-help' or ARGV.include? '--help' or ARGV.include? '--h'
-  STDERR.puts help_string
-  exit
-end
-
-default_pvalues = [0.0005]
-background = [1,1,1,1]
-rough_discretization = 1
-precise_discretization = 10
-output_file =  'collection.yaml'
-
-begin
-  folder = ARGV.shift
-  raise "No input. You'd specify folder with pat-files" unless folder
-  raise "Error! Folder #{folder} doesn't exist" unless Dir.exist?(folder)
-
-  pvalues = []
-  silent = false
-  until ARGV.empty?
-    case ARGV.shift
-      when '-b'
-        background = ARGV.shift(4).map(&:to_f)
-        raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless background == background.reverse
-      when '-p'
-        loop do
-          begin
-            Float(ARGV.first)
-            pvalues << ARGV.shift.to_f
-          rescue
-            raise StopIteration
-          end
-        end
-      when '-d'
-        rough_discretization, precise_discretization = ARGV.shift(2).map(&:to_f).sort
-      when '-o'
-        output_file = ARGV.shift
-      when '-m'
-        Macroape::MaxHashSizeSingle = ARGV.shift.to_f
-      when '-md'
-        Macroape::MaxHashSizeDouble = ARGV.shift.to_f
-      when '--silent'
-        silent = true
-      end
-  end
-  pvalues = default_pvalues if pvalues.empty?
-
-  Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
-  Macroape::MaxHashSizeDouble = 1000 unless defined? Macroape::MaxHashSizeDouble
-
-  collection = Macroape::Collection.new(rough_discretization, precise_discretization, background, pvalues)
-
-  current_dir = File.dirname(__FILE__)
-  Dir.glob(File.join(folder,'*')) do |filename|
-    STDERR.puts filename unless silent
-    pwm = Bioinform::PWM.new(File.read(filename))
-    pwm.name ||= File.basename(filename, File.extname(filename))
-
-    # When support of onefile collections is introduced - then here should be check if name exists.
-    # Otherwise it should skip motif and tell you about this
-    # Also two command line options to fail on skipping or to skip silently should be included
-
-    info = {rough: {}, precise: {}}
-    pwm.background(background)
-
-    pwm.discrete(rough_discretization).thresholds(*pvalues) do |pvalue, threshold, real_pvalue|
-      info[:rough][pvalue] = threshold / rough_discretization
-    end
-
-    pwm.discrete(precise_discretization).thresholds(*pvalues) do |pvalue, threshold, real_pvalue|
-      info[:precise][pvalue] = threshold / precise_discretization
-    end
-
-    collection.add_pwm(pwm, info)
-  end
-  File.open(output_file,'w') do |f|
-    f.puts(collection.to_yaml)
-  end
-rescue => err
-  STDERR.puts "\n#{err}\n#{err.backtrace.first(5).join("\n")}\n\nUse -help option for help\n"
-end