macroape 3.3.6 → 3.3.7

data/lib/macroape/cli/scan_collection.rb

@@ -1,4 +1,4 @@
-require 'macroape'
+require_relative '../../macroape'
 require 'yaml'
 
 module Macroape
@@ -31,7 +31,7 @@ module Macroape
           cat motifs/KLF4.pat | ruby scan_collection.rb .stdin collection.yaml -p 0.005 --precise 0.03
         }
 
-        if argv.empty? || ['-h', '--h', '-help', '--help'].any?{|help_option| argv.include?(help_option)}
+        if ['-h', '--h', '-help', '--help'].any?{|help_option| argv.include?(help_option)}
           STDERR.puts help_string
           exit
         end
@@ -90,34 +90,41 @@ module Macroape
         end
 
         query_pwm = data_model.new(query_input).to_pwm
-        query_pwm.background(background_query).max_hash_size(max_hash_size)
+        query_pwm.set_parameters(background: background_query, max_hash_size: max_hash_size)
         
         query_pwm_rough = query_pwm.discrete(collection.parameters.rough_discretization)
         query_pwm_precise = query_pwm.discrete(collection.parameters.precise_discretization)
 
-        query_threshold_rough = query_pwm_rough.threshold(pvalue)
-        query_threshold_precise = query_pwm_precise.threshold(pvalue)
+        query_threshold_rough, query_rough_real_pvalue = query_pwm_rough.threshold_and_real_pvalue(pvalue)        
+        query_threshold_precise, query_precise_real_pvalue = query_pwm_precise.threshold_and_real_pvalue(pvalue)
+        
+        if query_precise_real_pvalue == 0
+          $stderr.puts "Query motif #{query_pwm.name} gives 0 recognized words for a given P-value of #{pvalue} with the precise discretization level of #{collection.parameters.precise_discretization}. It's impossible to scan collection for this motif"
+          return
+        end
+        
+        if query_rough_real_pvalue == 0
+          query_pwm_rough, query_threshold_rough = query_pwm_precise, query_threshold_precise
+          $stderr.puts "Query motif #{query_pwm.name} gives 0 recognized words for a given P-value of #{pvalue} with the rough discretization level of #{collection.parameters.rough_discretization}. Forcing precise discretization level of #{collection.parameters.precise_discretization}"
+        end
 
         similarities = {}
         precision_file_mode = {}
 
-        collection.each do |pwm, pwm_info|
-          name = pwm.name
-          pwm.background(collection.parameters.background).max_hash_size(max_hash_size)
-          pwm_rough = pwm.discrete(collection.parameters.rough_discretization)
-          pwm_precise = pwm.discrete(collection.parameters.precise_discretization)
-
-          pwm_threshold_rough = pwm_info.rough[pvalue] * collection.parameters.rough_discretization
-          pwm_threshold_precise = pwm_info.precise[pvalue] * collection.parameters.precise_discretization
-
+        collection.each do |collection_pwm, pwm_info|
+          name = collection_pwm.name
           STDERR.puts name unless silent
-          cmp = Macroape::PWMCompare.new(query_pwm_rough, pwm_rough).max_hash_size(max_pair_hash_size)
-          info = cmp.jaccard(query_threshold_rough, pwm_threshold_rough)
-          precision_file_mode[name] = :rough
-
-          if precision_mode == :precise and info[:similarity] >= minimal_similarity
-            cmp = Macroape::PWMCompare.new(query_pwm_precise, pwm_precise).max_hash_size(max_pair_hash_size)
-            info = cmp.jaccard(query_threshold_precise, pwm_threshold_precise)
+          collection_pwm.set_parameters(background: collection.parameters.background, max_hash_size: max_hash_size)
+          if pwm_info.rough
+            collection_pwm_rough = collection_pwm.discrete(collection.parameters.rough_discretization)
+            collection_threshold_rough = pwm_info.rough[pvalue] * collection.parameters.rough_discretization
+            info = Macroape::PWMCompare.new(query_pwm_rough, collection_pwm_rough).set_parameters(max_pair_hash_size: max_pair_hash_size).jaccard(query_threshold_rough, collection_threshold_rough)
+            precision_file_mode[name] = :rough
+          end
+          if !pwm_info.rough || (precision_mode == :precise) && (info[:similarity] >= minimal_similarity)
+            collection_pwm_precise = collection_pwm.discrete(collection.parameters.precise_discretization)
+            collection_threshold_precise = pwm_info.precise[pvalue] * collection.parameters.precise_discretization
+            info = Macroape::PWMCompare.new(query_pwm_precise, collection_pwm_precise).set_parameters(max_pair_hash_size: max_pair_hash_size).jaccard(query_threshold_precise, collection_threshold_precise)
             precision_file_mode[name] = :precise
           end
           similarities[name] = info

data/lib/macroape/counting.rb

@@ -1,22 +1,16 @@
+require 'bioinform'
+
 module Bioinform
   class PWM
     # sets or gets limit size of calculation hash. It's a defence against overuse CPU resources by non-appropriate data
-    def max_hash_size!(new_max_hash_size)
-      @max_hash_size = new_max_hash_size
-      self
-    end
-    
-    def max_hash_size(*args)
-      case args.size
-      when 0 then @max_hash_size
-      when 1 then max_hash_size!(args.first)
-      else raise ArgumentError, '#max_hash_size method can get 0 or 1 argument'
-      end
-    end
+    make_parameters :max_hash_size
     
     def threshold(pvalue)
       thresholds(pvalue){|_, thresh, _| return thresh }
     end
+    def threshold_and_real_pvalue(pvalue)
+      thresholds(pvalue){|_, thresh, real_pv| return thresh, real_pv }
+    end
 
     def thresholds(*pvalues)
       thresholds_by_pvalues(*pvalues).each do |pvalue,(thresholds, counts)|
@@ -80,7 +74,7 @@ module Bioinform
         4.times do |letter|
           new_score = score + column[letter]
           if new_score >= least_sufficient
-            new_scores[new_score] += count * @background[letter]
+            new_scores[new_score] += count * background[letter]
           end
         end
       end

data/lib/macroape/pwm_compare.rb

@@ -1,21 +1,14 @@
+require 'bioinform/support/parameters'
+
 module Macroape
   class PWMCompare
+    include Parameters
     # sets or gets limit of summary size of calculation hash. It's a defence against overuse CPU resources by non-appropriate data
-    def max_hash_size!(new_max_hash_size)
-      @max_hash_size = new_max_hash_size
-      self
-    end
-    
-    def max_hash_size(*args)
-      case args.size
-      when 0 then @max_hash_size
-      when 1 then max_hash_size!(args.first)
-      else raise ArgumentError, '#max_hash_size method can get 0 or 1 argument'
-      end
-    end
-  
-    attr_reader :first, :second
+    make_parameters :max_pair_hash_size
+
+    attr_reader :first, :second, :parameters
     def initialize(first, second)
+      @parameters = OpenStruct.new
       @first = first
       @second = second
     end
@@ -34,7 +27,7 @@ module Macroape
 
     def each_alignment
       (-second.length..first.length).to_a.product([:direct,:revcomp]) do |shift, orientation|
-        yield PWMCompareAligned.new(first, second, shift, orientation).max_hash_size(max_hash_size)
+        yield PWMCompareAligned.new(first, second, shift, orientation).set_parameters(max_pair_hash_size: max_pair_hash_size)
       end
     end
 

data/lib/macroape/pwm_compare_aligned.rb

@@ -1,9 +1,16 @@
-require 'macroape/aligned_pair_intersection'
+require 'bioinform/support/parameters'
+require_relative './aligned_pair_intersection'
 
 module Macroape
   class PWMCompareAligned
-    attr_reader :first, :second, :length, :shift, :orientation, :first_length, :second_length
+    include Parameters
+    # sets or gets limit of summary size of calculation hash. It's a defence against overuse CPU resources by non-appropriate data
+    make_parameters :max_pair_hash_size
+
+    attr_reader :first, :second, :length, :shift, :orientation, :first_length, :second_length, :parameters
+    
     def initialize(first_unaligned, second_unaligned, shift, orientation)
+      @parameters = OpenStruct.new
       @shift, @orientation = shift, orientation
 
       @first_length, @second_length = first_unaligned.length, second_unaligned.length
@@ -21,6 +28,8 @@ module Macroape
       @first = first.right_augment(@length - first.length)
       @second = second.right_augment(@length - second.length)
     end
+    
+    
 
     def direct?
       orientation == :direct
@@ -80,26 +89,6 @@ module Macroape
       end
     end
 
-=begin
-    def discrete(rate)
-      PWMCompareAligned.new(first.discrete(rate), second.discrete(rate))
-    end
-
-    def sort_pair_of_matrices_by(&block)
-      mat = first.pwm.zip(second.pwm).sort_by(&block).transpose
-      PWMCompareAligned.new(SinglePWM(mat[0],first.probabilities), SinglePWM(mat[1], second.probabilities))
-    end
-    def sort_decreasing_max
-      PWMCompareAligned.new(*sort_pair_of_matrices_by{|col_pair| -col_pair[0].max} )
-    end
-    def sort_increasing_min
-      PWMCompareAligned.new(*sort_pair_of_matrices_by{|col_pair| col_pair[0].min} )
-    end
-    def permute_columns(permutation_index)
-      PWMCompareAligned.new(first.permute(permutation_index), second.permute(permutation_index))
-    end
-=end
-
     def jaccard(first_threshold, second_threshold)
       f = first.counts_by_thresholds(first_threshold).first
       s = second.counts_by_thresholds(second_threshold).first
@@ -131,22 +120,6 @@ module Macroape
         [first_len - shift, second_len].max
       end
     end
-    
-    # sets or gets limit of summary size of calculation hash. It's a defence against overuse CPU resources by non-appropriate data
-    def max_hash_size!(new_max_hash_size)
-      @max_hash_size = new_max_hash_size
-      self
-    end
-    
-    def max_hash_size(*args)
-      case args.size
-      when 0 then @max_hash_size
-      when 1 then max_hash_size!(args.first)
-      else raise ArgumentError, '#max_hash_size method can get 0 or 1 argument'
-      end
-    end
-    
-
   end
 
 end

data/lib/macroape/version.rb

@@ -1,3 +1,3 @@
 module Macroape
-  VERSION = "3.3.6"
+  VERSION = "3.3.7"
 end

data/lib/macroape.rb

@@ -1,11 +1,10 @@
-require 'macroape/version'
+require_relative 'macroape/version'
 
-require 'bioinform'
-require 'macroape/counting'
-
-require 'macroape/aligned_pair_intersection'
-require 'macroape/pwm_compare_aligned'
-require 'macroape/pwm_compare'
+require_relative 'macroape/counting'
+require_relative 'macroape/aligned_pair_intersection'
+require_relative 'macroape/pwm_compare_aligned'
+require_relative 'macroape/pwm_compare'
+require_relative 'macroape/cli'
 
 module Macroape
   # Your code goes here...

data/macroape.gemspec

@@ -15,5 +15,5 @@ Gem::Specification.new do |gem|
   gem.require_paths = ["lib"]
   gem.version       = Macroape::VERSION
   
-  gem.add_dependency('bioinform', '>= 0.1.7')
+  gem.add_dependency('bioinform', '= 0.1.8')
 end

data/spec/count_distribution_spec.rb

@@ -1,6 +1,5 @@
-require 'spec_helper'
-require 'bioinform'
-require 'macroape/counting'
+require_relative 'spec_helper'
+require_relative '../lib/macroape/counting'
 
 describe Bioinform::PWM do
   let :matrix_first do [[1,2,3,4],[10,20,30,40],[100,200,300,400]] end

data/spec/spec_helper.rb

@@ -1,4 +1,4 @@
-$lib_folder = File.dirname(__FILE__) + '/../lib'
-$LOAD_PATH.unshift $lib_folder
+$bioinform_folder = File.expand_path(File.join(File.dirname(__FILE__), '..', '..', 'bioinform', 'lib'))
+$LOAD_PATH.unshift $bioinform_folder
 
 require 'rspec'

data/test/align_motifs_test.rb

@@ -1,4 +1,4 @@
-require 'test_helper'
+require_relative 'test_helper'
 
 class TestAlignmotifs < Test::Unit::TestCase
   def test_align_motifs

data/test/data/KLF4_f2_scan_results_all.txt

@@ -1,4 +1,4 @@
-#pwm	similarity	shift	overlap	orientation
-KLF4_f2	1.0	0	10	direct
-SP1_f1	0.22754919499105544	-1	10	direct
-GABPA_f1	0.00043527658136684877	-8	5	direct
+#pwm	similarity	shift	overlap	orientation
+KLF4_f2	1.0	0	10	direct
+SP1_f1	0.22754919499105544	-1	10	direct
+GABPA_f1	0.00043527658136684877	-8	5	direct

data/test/data/KLF4_f2_scan_results_precise_mode.txt

@@ -1,4 +1,4 @@
-#pwm	similarity	shift	overlap	orientation
-KLF4_f2	1.0	0	10	direct	*
-SP1_f1	0.2420758234928527	-1	10	direct	*
-GABPA_f1	0.00043527658136684877	-8	5	direct
+#pwm	similarity	shift	overlap	orientation
+KLF4_f2	1.0	0	10	direct	*
+SP1_f1	0.2420758234928527	-1	10	direct	*
+GABPA_f1	0.00043527658136684877	-8	5	direct

data/test/data/collection_pcm_without_thresholds.yaml

@@ -0,0 +1,185 @@
+--- &18345276 !ruby/object:Bioinform::Collection
+collection:
+- - !ruby/object:Bioinform::PCM
+    parameters: !ruby/object:OpenStruct
+      table:
+        :name: GABPA_f1
+        :tags:
+        - *18345276
+        :background:
+        - 1
+        - 1
+        - 1
+        - 1
+      modifiable: true
+    matrix:
+    - - 615.2572649050138
+      - 697.0698715160123
+      - 1261.1903440712872
+      - 176.43506582414153
+    - - 996.4929869323321
+      - 805.1878697364007
+      - 693.7695793644275
+      - 254.5021102832924
+    - - 1106.9888035794224
+      - 508.19444415177276
+      - 1029.8329748714536
+      - 104.93632371380718
+    - - 143.7121486195701
+      - 2086.4279160661263
+      - 518.37507049306
+      - 1.4374111377025893
+    - - 362.9541452731307
+      - 2369.473894845734
+      - 17.23702397004065
+      - 0.2874822275405179
+    - - 0.0
+      - 0.0
+      - 2749.952546316428
+      - 0.0
+    - - 0.0
+      - 0.0
+      - 2749.952546316428
+      - 0.0
+    - - 2748.2567506938462
+      - 1.695795622582083
+      - 0.0
+      - 0.0
+    - - 2726.6484322711017
+      - 1.1499289101620715
+      - 1.1499289101620715
+      - 21.00425622500253
+    - - 202.05697400573305
+      - 28.799402471063658
+      - 2518.808687612104
+      - 0.2874822275405179
+    - - 172.92889618879767
+      - 521.1240363384483
+      - 106.38197600987633
+      - 1949.517637779338
+    - - 398.1679460365911
+      - 424.20938204069563
+      - 1706.4024212088275
+      - 221.17279703034018
+    - - 764.2587933951809
+      - 675.0883944902433
+      - 1066.5413633225007
+      - 244.06399510852864
+  - !ruby/object:OpenStruct
+    table: {}
+- - !ruby/object:Bioinform::PCM
+    parameters: !ruby/object:OpenStruct
+      table:
+        :name: KLF4_f2
+        :tags:
+        - *18345276
+        :background:
+        - 1
+        - 1
+        - 1
+        - 1
+      modifiable: true
+    matrix:
+    - - 1233.46088405354
+      - 93.18173277811673
+      - 1036.6014857092885
+      - 1258.2948629970272
+    - - 263.979242343185
+      - 5.314520555872139
+      - 3347.5949971525274
+      - 4.650205486388122
+    - - 76.7700780003465
+      - 6.643150694840173
+      - 3529.4896409394937
+      - 8.636095903292224
+    - - 57.86097393406657
+      - 18.102585643439472
+      - 3520.3342027139347
+      - 25.24120324653207
+    - - 518.1947904009378
+      - 1545.9062946905135
+      - 22.396758181071043
+      - 1535.0411222654507
+    - - 137.98151691820345
+      - 9.300410972776241
+      - 3456.320530770924
+      - 17.936506876068467
+    - - 115.27647661640499
+      - 81.51802997128804
+      - 1861.9425868567278
+      - 1562.801872093553
+    - - 227.8095486111286
+      - 42.84555258785854
+      - 3278.6396005325996
+      - 72.244263806387
+    - - 108.73384179997886
+      - 134.47328134862394
+      - 3162.880454846513
+      - 215.45138754285665
+    - - 238.49636899561344
+      - 2225.9561104691043
+      - 402.40727964384774
+      - 754.6792064294074
+  - !ruby/object:OpenStruct
+    table: {}
+- - !ruby/object:Bioinform::PCM
+    parameters: !ruby/object:OpenStruct
+      table:
+        :name: SP1_f1
+        :tags:
+        - *18345276
+        :background:
+        - 1
+        - 1
+        - 1
+        - 1
+      modifiable: true
+    matrix:
+    - - 682.6436366358055
+      - 443.1455214015781
+      - 2075.655346294993
+      - 287.211468117951
+    - - 299.8883246804867
+      - 103.74338315843572
+      - 2613.8927022405364
+      - 471.1315623708902
+    - - 591.4892493324709
+      - 42.631827541794564
+      - 2845.1654083148564
+      - 9.36948726124641
+    - - 7.071084742361592
+      - 45.29093411231232
+      - 3432.8847704374107
+      - 3.409183158303573
+    - - 91.308984085713
+      - 19.1536481364332
+      - 3373.656949880137
+      - 4.5363903481026
+    - - 809.2082973387932
+      - 2246.941954176211
+      - 61.30766021687515
+      - 371.19806071846244
+    - - 120.56476435866055
+      - 42.4349244403591
+      - 3242.1560628684038
+      - 83.50022078295852
+    - - 13.72524477409959
+      - 35.858220519297525
+      - 3332.4066864946167
+      - 106.66582066236779
+    - - 558.1188080161639
+      - 90.0084504200356
+      - 2694.854973210736
+      - 145.67374080342415
+    - - 264.0088462230318
+      - 254.7175868081866
+      - 2796.88087480315
+      - 173.0486646159857
+    - - 519.46013914282
+      - 1874.9349086474765
+      - 654.5411208373813
+      - 439.7198038226514
+  - !ruby/object:OpenStruct
+    table: {}
+parameters: !ruby/object:OpenStruct
+  table: {}

data/test/data/collection_without_thresholds.yaml

@@ -0,0 +1,185 @@
+--- &18740484 !ruby/object:Bioinform::Collection
+collection:
+- - !ruby/object:Bioinform::PWM
+    parameters: !ruby/object:OpenStruct
+      table:
+        :name: GABPA_f1
+        :tags:
+        - *18740484
+        :background:
+        - 1
+        - 1
+        - 1
+        - 1
+      modifiable: true
+    matrix:
+    - - -0.1106670158341858
+      - 0.013801606113892391
+      - 0.6054596108973699
+      - -1.3518085041421573
+    - - 0.37030668921643345
+      - 0.15761121480429963
+      - 0.009069314183831202
+      - -0.9888619717703562
+    - - 0.47526546359546684
+      - -0.3011678534572083
+      - 0.4031522994412777
+      - -1.8638752827041059
+    - - -1.5544255540164373
+      - 1.1082369687811506
+      - -0.2814091552834454
+      - -5.30708531823271
+    - - -0.6362037835776368
+      - 1.235338189985594
+      - -3.5801322928552253
+      - -5.717323067092849
+    - - -5.852906870733575
+      - -5.852906870733575
+      - 1.3841383838057746
+      - -5.852906870733575
+    - - -5.852906870733575
+      - -5.852906870733575
+      - 1.3841383838057746
+      - -5.852906870733575
+    - - 1.3835219739184708
+      - -5.2341956006430985
+      - -5.852906870733575
+      - -5.852906870733575
+    - - 1.3756340514956562
+      - -5.394962755562375
+      - -5.394962755562375
+      - -3.401117964959733
+    - - -1.2176198315414444
+      - -3.109079898175411
+      - 1.2964067931472216
+      - -5.717323067092849
+    - - -1.3716559438167257
+      - -0.2761401935045069
+      - -1.8504445165866068
+      - 1.0404320473626856
+    - - -0.5440863133031895
+      - -0.48103682561971345
+      - 0.907381908447086
+      - -1.1280642594012078
+    - - 0.10557340209290218
+      - -0.01814819455289191
+      - 0.4381106695354074
+      - -1.0304105539540915
+  - !ruby/object:OpenStruct
+    table: {}
+- - !ruby/object:Bioinform::PWM
+    parameters: !ruby/object:OpenStruct
+      table:
+        :name: KLF4_f2
+        :tags:
+        - *18740484
+        :background:
+        - 1
+        - 1
+        - 1
+        - 1
+      modifiable: true
+    matrix:
+    - - 0.30861857265872605
+      - -2.254321000121579
+      - 0.13505703522674192
+      - 0.3285194224375633
+    - - -1.227018967707036
+      - -4.814127713368663
+      - 1.3059890687390967
+      - -4.908681463544344
+    - - -2.443469374521196
+      - -4.648238485031404
+      - 1.3588686548279805
+      - -4.441801801188402
+    - - -2.7177827948276123
+      - -3.8073538975356565
+      - 1.356272809724262
+      - -3.504104725510225
+    - - -0.5563232977367343
+      - 0.5340697765121405
+      - -3.61417723090579
+      - 0.5270259776377405
+    - - -1.8687622060887386
+      - -4.381483976582316
+      - 1.337932245336098
+      - -3.815629658877517
+    - - -2.045671123823928
+      - -2.384975142213679
+      - 0.7198551207724355
+      - 0.5449254135616948
+    - - -1.373157530374372
+      - -3.0063112097748217
+      - 1.285188335493552
+      - -2.5026044231773543
+    - - -2.1030513122772208
+      - -1.8941348100402244
+      - 1.249265758393991
+      - -1.4284210948906104
+    - - -1.3277128628152939
+      - 0.8982415633049462
+      - -0.8080773665408135
+      - -0.18161647647456935
+  - !ruby/object:OpenStruct
+    table: {}
+- - !ruby/object:Bioinform::PWM
+    parameters: !ruby/object:OpenStruct
+      table:
+        :name: SP1_f1
+        :tags:
+        - *18740484
+        :background:
+        - 1
+        - 1
+        - 1
+        - 1
+      modifiable: true
+    matrix:
+    - - -0.24435707885585292
+      - -0.674823404693731
+      - 0.8657012535789866
+      - -1.1060188862599287
+    - - -1.0631255752097797
+      - -2.111925969423868
+      - 1.0960627561110403
+      - -0.6138563775211977
+    - - -0.3872276234760535
+      - -2.9739851913218045
+      - 1.1807800242010378
+      - -4.338927525031566
+    - - -4.563896055436894
+      - -2.9161633002532277
+      - 1.3684371349982638
+      - -5.077972423609655
+    - - -2.2369752892820083
+      - -3.7196436313301846
+      - 1.3510439136452734
+      - -4.889930670508233
+    - - -0.07473964149330865
+      - 0.944919654762011
+      - -2.6246857648086044
+      - -0.8510983487822436
+    - - -1.9643526491643322
+      - -2.978402770880115
+      - 1.3113096718240573
+      - -2.324334259499025
+    - - -4.0155484139655835
+      - -3.1384268078096667
+      - 1.3387488589788057
+      - -2.084673903537648
+    - - -0.44509385828355363
+      - -2.2510053061629702
+      - 1.1265431574368685
+      - -1.7780413702431372
+    - - -1.1896356092245048
+      - -1.2251832285630027
+      - 1.1636760063747527
+      - -1.6080243648157353
+    - - -0.5166047365590571
+      - 0.7641033353626657
+      - -0.2862677570028208
+      - -0.68254820978656
+  - !ruby/object:OpenStruct
+    table: {}
+parameters: !ruby/object:OpenStruct
+  table: {}

data/test/data/medium_motif.pat

@@ -0,0 +1,8 @@
+medium_motif_name
+0.30861857265872605 -2.254321000121579 0.13505703522674192 0.3285194224375633 
+-1.227018967707036 -4.814127713368663 1.3059890687390967 -4.908681463544344 
+-2.443469374521196 -4.648238485031404 1.3588686548279805 -4.441801801188402 
+-2.7177827948276123 -3.8073538975356565 1.356272809724262 -3.504104725510225 
+-0.5563232977367343 0.5340697765121405 -3.61417723090579 0.5270259776377405 
+-1.8687622060887386 -4.381483976582316 1.337932245336098 -3.815629658877517 
+-2.045671123823928 -2.384975142213679 0.7198551207724355 0.5449254135616948