macroape 3.3.3 → 3.3.4

data/lib/macroape/exec/scan_collection.rb

@@ -1,124 +0,0 @@
-help_string = %q{
-Command-line format:
-ruby scan_collection.rb <pat-file> <collection> [options]
-        or in linux
-cat <pat-file> | ruby scan_collection.rb .stdin <collection> [options]
-        or on windows
-type <pat-file> | ruby scan_collection.rb .stdin <collection> [options]
-
-Options:
-  [-p <P-value>]
-  [-c <similarity cutoff (minimal similarity to be included in output)> ] or [--all], '-c 0.05' by default
-  [--precise [<level, minimal similarity to check on a more precise discretization level on the second pass>]], off by default, '--precise 0.01' if level is not set
-  [--silent] - don't show current progress information during scan (by default this information's written into stderr)
-
-Output format:
- <name> <similarity jaccard index> <shift> <overlap> <orientation> * [in case that result calculated on the second pass(in precise mode)]
-    Attention! Name can contain whitespace characters.
-    Attention! The shift and orientation are reported for the collection matrix relative to the query matrix.
-
-Example:
-  ruby scan_collection.rb motifs/KLF4.pat collection.yaml -p 0.005
-            or in linux
-  cat motifs/KLF4.pat | ruby scan_collection.rb .stdin collection.yaml -p 0.005 --precise 0.03
-}
-
-$:.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
-
-begin
-  filename = ARGV.shift
-  collection_file = ARGV.shift
-  raise "No input. You'd specify input source for pat: filename or .stdin" unless filename
-  raise "No input. You'd specify input file with collection" unless collection_file
-  raise "Collection file #{collection_file} doesn't exist" unless File.exist?(collection_file)
-
-  pvalue = 0.0005
-  cutoff = 0.05 # minimal similarity to output
-  collection = YAML.load_file(collection_file)
-  background_query = collection.background
-
-  silent = false
-  precision_mode = :rough
-  until ARGV.empty?
-    case ARGV.shift
-      when '-bq'
-        background_query = ARGV.shift(4).map(&:to_f)
-        raise 'background should be symmetric: p(A)=p(T) and p(G) = p(C)' unless background_query == background_query.reverse
-      when '-p'
-        pvalue = ARGV.shift.to_f
-      when '-m'
-        Macroape::MaxHashSizeSingle = ARGV.shift.to_f
-      when '-md'
-        Macroape::MaxHashSizeDouble = ARGV.shift.to_f
-      when '-c'
-        cutoff = ARGV.shift.to_f
-      when '--all'
-        cutoff = 0.0
-      when '--silent'
-        silent = true
-      when '--precise'
-        precision_mode = :precise
-        begin
-          Float(ARGV.first)
-          minimal_similarity = ARGV.shift.to_f
-        rescue
-          minimal_similarity = 0.05
-        end
-    end
-  end
-  Macroape::MaxHashSizeSingle = 1000000 unless defined? Macroape::MaxHashSizeSingle
-  Macroape::MaxHashSizeDouble = 1000 unless defined? Macroape::MaxHashSizeDouble
-
-  raise "Thresholds for pvalue #{pvalue} aren't presented in collection (#{collection.pvalues.join(', ')}). Use one of listed pvalues or recalculate the collection with needed pvalue" unless collection.pvalues.include? pvalue
-
-  if filename == '.stdin'
-#    query_pwm = Macroape::SingleMatrix.load_from_stdin(STDIN)
-  else
-    raise "Error! File #{filename} doesn't exist" unless File.exist?(filename)
-    query_pwm = Bioinform::PWM.new(File.read(filename))
-  end
-
-
-  query_pwm_rough = query_pwm.background(background_query).discrete(collection.rough_discretization)
-  query_pwm_precise = query_pwm.background(background_query).discrete(collection.precise_discretization)
-
-  threshold = query_pwm_rough.threshold(pvalue)
-  threshold_precise = query_pwm_precise.threshold(pvalue)
-
-  similarities = {}
-  precision_file_mode = {}
-
-  collection.pwms.each_key do |name|
-    pwm = collection.pwms[name]
-    pwm_info = collection.infos[name]
-    STDERR.puts pwm.name unless silent
-    cmp = Macroape::PWMCompare.new(query_pwm_rough, pwm.background(collection.background).discrete(collection.rough_discretization))
-    info = cmp.jaccard(threshold, pwm_info[:rough][pvalue] * collection.rough_discretization)
-    precision_file_mode[name] = :rough
-
-    if precision_mode == :precise and info[:similarity] >= minimal_similarity
-      cmp = Macroape::PWMCompare.new(query_pwm_precise, pwm.background(collection.background).discrete(collection.precise_discretization))
-      info = cmp.jaccard(threshold_precise, pwm_info[:precise][pvalue] * collection.precise_discretization)
-      precision_file_mode[name] = :precise
-    end
-    similarities[name] = info
-  end
-
-  puts "#pwm\tsimilarity\tshift\toverlap\torientation"
-  similarities.sort_by do |name, info|
-    info[:similarity]
-  end.reverse.each do |name, info|
-    precision_text = (precision_file_mode[name] == :precise) ? "\t*" : ""
-    puts "#{name}\t#{info[:similarity]}\t#{info[:shift]}\t#{info[:overlap]}\t#{info[:orientation]}#{precision_text}" if info[:similarity] >= cutoff
-  end
-
-rescue => err
-  STDERR.puts "\n#{err}\n#{err.backtrace.first(5).join("\n")}\n\nUse -help option for help\n"
-end

data/test/eval_alignment_similarity_test.rb

@@ -1,20 +0,0 @@
-require 'test_helper'
-
-puts "\n\neval_alignment test:"
-class TestEvalAlignmentSimilarity < Test::Unit::TestCase
-  def test_process_at_optimal_alignment
-    IO.popen(Helpers.exec_cmd('eval_alignment','test/data/KLF4_f2.pat test/data/SP1_f1.pat -1 direct')){|f|
-      assert_equal "0.2420758234928527\n779.0\t11\n.>>>>>>>>>>\n>>>>>>>>>>>\n-1\tdirect\n", f.read
-    }
-  end
-  def test_process_not_optimal_alignment
-    IO.popen(Helpers.exec_cmd('eval_alignment','test/data/KLF4_f2.pat test/data/SP1_f1.pat 0 direct')){|f|
-      assert_equal "0.0017543859649122807\n7.0\t11\n>>>>>>>>>>.\n>>>>>>>>>>>\n0\tdirect\n", f.read
-    }
-  end
-  def test_process_at_optimal_alignment_reversed
-    IO.popen(Helpers.exec_cmd('eval_alignment','test/data/KLF4_f2.pat test/data/SP1_f1.pat -1 revcomp')){|f|
-      assert_equal "0.0\n0.0\t11\n.>>>>>>>>>>\n<<<<<<<<<<<\n-1\trevcomp\n", f.read
-    }
-  end
-end