bioinform 0.1.1 → 0.1.2

data/lib/bioinform/data_models/parser.rb

@@ -2,40 +2,37 @@ require 'bioinform/support'
 require 'bioinform/data_models/pm'
 
 module Bioinform
-  class PM
-    class Parser
-      attr_reader :input
-      
-      @subclasses ||= []
-      class << self
-        def subclasses
-          @subclasses
-        end  
-        def inherited(subclass)
-          PM::Parser.subclasses << subclass
-        end
-      end
-      
-      def initialize(input)
-        @input = input
-      end
-      
-      def parse_core
-        raise NotImplemented
-      end
-      
-        
-      def parse
-        parse_core
+  class Parser
+    attr_reader :input
+    
+    @subclasses ||= []
+    class << self
+      def subclasses
+        @subclasses
+      end  
+      def inherited(subclass)
+        Parser.subclasses << subclass
       end
+    end
+    
+    def initialize(input)
+      @input = input
+    end
+    
+    def parse_core
+      raise NotImplemented
+    end
+    
       
-      def can_parse?
-        parse_core
-        true
-      rescue
-        false
-      end
-
+    def parse
+      parse_core
+    end
+    
+    def can_parse?
+      parse_core
+      true
+    rescue
+      false
     end
   end
 end

data/lib/bioinform/data_models/parsers/array_parser.rb

@@ -2,7 +2,7 @@ require 'bioinform/support'
 require 'bioinform/data_models/parser'
 
 module Bioinform
-  class ArrayParser < PM::Parser
+  class ArrayParser < Parser
     def parse_core
       raise ArgumentError  unless input.is_a?(Array) && input.all?(&:is_a?.(Array)) && input.same_by?(&:length) && (input.size == 4 || input.sample.size == 4)
       if input.all?{|line| line.size == 4}

data/lib/bioinform/data_models/parsers/hash_parser.rb

@@ -2,7 +2,7 @@ require 'bioinform/support'
 require 'bioinform/data_models/parser'
 
 module Bioinform
-  class HashParser < PM::Parser
+  class HashParser < Parser
     def parse_core
       case input
       when Hash

data/lib/bioinform/data_models/parsers/string_parser.rb

@@ -3,15 +3,15 @@ require 'bioinform/data_models/parser'
 require 'bioinform/data_models/parsers/array_parser'
 
 module Bioinform
-  class StringParser < PM::Parser    
+  class StringParser < Parser
     def number_pat
-      '[+-]?\d+(\.\d+)?'
+      '[+-]?\d+(\.\d+)?([eE][+-]?\d{1,3})?'
     end
     def row_pat 
       "(#{number_pat} )*#{number_pat}"
     end
     def name_pat
-      '>? ?(?<name>[\w.+:-]+)'
+      '(>\s*)?(?<name>\S+)'
     end
     def matrix_pat 
       "(?<matrix>(#{row_pat}\n)*#{row_pat})"

data/lib/bioinform/data_models/parsers.rb

@@ -1,3 +1,5 @@
+# Parsers should be included after PM class defined - in order it could catch definitions of new classes
+
 require 'bioinform/data_models/parsers/array_parser'
 require 'bioinform/data_models/parsers/hash_parser'
 require 'bioinform/data_models/parsers/string_parser'

data/lib/bioinform/data_models/pm.rb

@@ -1,6 +1,9 @@
 require 'bioinform/support'
 
 module Bioinform
+  IndexByLetter = {'A' => 0, 'C' => 1, 'G' => 2, 'T' => 3}
+  LetterByIndex = {0 => 'A', 1 => 'C', 2 => 'G', 3 => 'T'}
+
   class PM
     attr_reader :matrix, :background
     attr_accessor :name
@@ -22,7 +25,7 @@ module Bioinform
       if @parser
         raise ArgumentError, 'Input cannot be parsed by specified parser' unless @parser.new(@input).can_parse?
       else
-        @parser = PM::Parser.subclasses.find{|parser_class| parser_class.new(@input).can_parse? }
+        @parser = Parser.subclasses.find{|parser_class| parser_class.new(@input).can_parse? }
         raise ArgumentError, 'No one parser can parse specified input' unless @parser
       end
     end

data/lib/bioinform/data_models/pwm.rb

@@ -1,12 +1,10 @@
 require 'bioinform/support'
 require 'bioinform/data_models/pm'
 module Bioinform
-
   class PWM < PM
     def score_mean
       matrix.inject(0.0){ |mean, position| mean + position.each_index.inject(0.0){|sum, letter| sum + position[letter] * probability[letter]} }
     end
-    
     def score_variance
       matrix.inject(0.0) do |variance, position|
         variance  + position.each_index.inject(0.0) { |sum,letter| sum + position[letter]**2 * probability[letter] } -
@@ -19,5 +17,12 @@ module Bioinform
       n_ = Math.inverf(1 - 2 * pvalue) * Math.sqrt(2)
       score_mean + n_ * sigma
     end
+    
+    def score(word)
+      word = word.upcase
+      raise ArgumentError  unless word.length == length
+      raise ArgumentError  unless word.each_char.all?{|letter| %w{A C G T}.include? letter}
+      word.each_char.map.with_index{|letter, pos| matrix[pos][IndexByLetter[letter]] }.inject(0.0, &:+)
+    end
   end
 end

data/lib/bioinform/data_models.rb

@@ -1,6 +1,6 @@
 require 'bioinform/data_models/pm'
-#require 'bioinform/data_models/pcm'
+require 'bioinform/data_models/pcm'
 require 'bioinform/data_models/pwm'
-#require 'bioinform/data_models/ppm'
+require 'bioinform/data_models/ppm'
 #require 'bioinform/data_models/iupac_word'
 #require 'bioinform/data_models/iupac_wordset'

data/lib/bioinform/support/callable_symbol.rb

@@ -1,5 +1,3 @@
-require 'bioinform/support/curry_except_self'
-
 # Useful extension for &:symbol - syntax to make it possible to pass arguments for method in block
 #   ['abc','','','def','ghi'].tap(&:delete.(''))             # ==> ['abc','def','ghi']
 #   [1,2,3].map(&:to_s.(2))                                  # ==> ['1','10','11']
@@ -8,7 +6,7 @@ require 'bioinform/support/curry_except_self'
 #   [%w{1 2 3 4 5},%w{6 7 8 9}].map(&:join.().length)         # ==> [5,4]
 class Symbol
   def call(*args, &block)
-    obj=BasicObject.new.instance_exec(self,args,block) do |meth,params,block| 
+    obj = BasicObject.new.instance_exec(self,args,block) do |meth,params,block| 
       @postprocess_meth = [meth]
       @postprocess_args = [params]
       @postprocess_block = [block]
@@ -37,4 +35,16 @@ class Symbol
     
     obj
   end
-end
+end
+
+
+=begin
+# Much simplier but ['abc','cdef','xy','z','wwww'].select(&:size.() == 4) wouldn't work
+class Symbol
+  def call(*args, &block)
+    proc do |recv|
+      recv.__send__(self, *args, &block)
+    end
+  end
+end
+=end

data/lib/bioinform/support.rb

@@ -3,7 +3,6 @@ require 'active_support/core_ext/hash/indifferent_access'
 
 require 'bioinform/support/callable_symbol'
 require 'bioinform/support/collect_hash'
-require 'bioinform/support/curry_except_self'
 require 'bioinform/support/delete_many'
 require 'bioinform/support/has_keys'
 require 'bioinform/support/multiline_squish'

data/lib/bioinform/version.rb

@@ -1,3 +1,3 @@
 module Bioinform
-  VERSION = "0.1.1"
+  VERSION = "0.1.2"
 end

data/spec/data_models/parser_spec.rb

@@ -2,8 +2,8 @@ require 'spec_helper'
 require 'bioinform/data_models/parser'
 
 module Bioinform
-  describe PM::Parser do
-    include PM::Parser::Helpers
+  describe Parser do
+    include Parser::Helpers
     
     before :each do
       parser_stub :ParserBad, false, { matrix: [[0,0,0,0],[1,1,1,1]], name: 'Bad' }
@@ -14,9 +14,9 @@ module Bioinform
     end
 
     context 'when subklass created' do
-      it 'PM::Parser.subclasses should contain all subclasses' do
-        PM::Parser.subclasses.should include ParserBad
-        PM::Parser.subclasses.should include ParserGood
+      it 'Parser.subclasses should contain all subclasses' do
+        Parser.subclasses.should include ParserBad
+        Parser.subclasses.should include ParserGood
       end
     end
     
@@ -29,7 +29,7 @@ module Bioinform
     
     describe '#parse' do
       it 'should raise an error unless reimplemented' do
-        parser = PM::Parser.new('my stub input')
+        parser = Parser.new('my stub input')
         expect{ parser.parse }.to raise_error
       end
       

data/spec/data_models/parsers/string_parser_spec.rb

@@ -48,6 +48,31 @@ module Bioinform
         9 -8.7 6.54 -3210
       EOS
       
+      @input_with_exponent = <<-EOS
+        1.23 4.56 7.8 9.0
+        9 -8.7 6.54 -3.210e3
+      EOS
+      
+      @input_with_plus_exponent = <<-EOS
+        1.23 4.56 7.8 9.0
+        9 -8.7 6.54 -3.210e+3
+      EOS
+      
+      @input_with_minus_exponent = <<-EOS
+        1.23 4.56 7.8 9.0
+        9 -87e-1 6.54 -3210
+      EOS
+      
+      @input_with_upcase_exponent = <<-EOS
+        1.23 4.56 7.8 9.0
+        9 -8.7 6.54 -3.210E3
+      EOS
+      
+      @input_with_manydigit_exponent = <<-EOS
+        1.23 4.56 7.8 9.0
+        9 -8.7 6.54 -0.0000003210e10
+      EOS
+      
       @input_transposed = <<-EOS
         1.23 9  
         4.56 -8.7
@@ -55,14 +80,15 @@ module Bioinform
         9.0 -3210
       EOS
       
+      
       @bad_input_not_numeric = <<-EOS
         1.23 4.56 aaa 9.0
         9 -8.7 6.54 -3210
       EOS
       
       @bad_input_different_row_size = <<-EOS
-        1.23 4.56 7.8
-        9 -8.7 6.54 -3210
+        1.23 4.56 7.8 9
+        9 -8.7 6.54
       EOS
       
       @bad_input_not_4_rows_and_cols = <<-EOS
@@ -72,6 +98,11 @@ module Bioinform
         4     5     6
         1     1     1
       EOS
+      
+      @bad_input_with_empty_exponent = <<-EOS
+        1.23 4.56 7.8 9.0
+        9e -8.7 6.54 3210
+      EOS
     end
     
     describe '#can_parse?' do
@@ -84,11 +115,18 @@ module Bioinform
         StringParser.new(@input_with_leading_and_finishing_spaces_and_newlines).can_parse?.should be_true
         StringParser.new(@input_without_name).can_parse?.should be_true
         StringParser.new(@input_transposed).can_parse?.should be_true
+        StringParser.new(@input_with_exponent).can_parse?.should be_true
+        StringParser.new(@input_with_plus_exponent).can_parse?.should be_true
+        StringParser.new(@input_with_minus_exponent).can_parse?.should be_true
+        StringParser.new(@input_with_upcase_exponent).can_parse?.should be_true
+        StringParser.new(@input_with_manydigit_exponent).can_parse?.should be_true
+        
       end
       it 'should return false for invalid input string' do
         StringParser.new(@bad_input_not_numeric).can_parse?.should be_false
         StringParser.new(@bad_input_different_row_size).can_parse?.should be_false
         StringParser.new(@bad_input_not_4_rows_and_cols).can_parse?.should be_false
+        StringParser.new(@bad_input_with_empty_exponent).can_parse?.should be_false
       end
     end
     describe '#parse' do
@@ -101,11 +139,17 @@ module Bioinform
         StringParser.new(@input_with_leading_and_finishing_spaces_and_newlines).parse.should == {matrix: @matrix}
         StringParser.new(@input_without_name).parse.should == {matrix: @matrix}
         StringParser.new(@input_transposed).parse.should == {matrix: @matrix}
+        StringParser.new(@input_with_exponent).parse.should == {matrix: @matrix}
+        StringParser.new(@input_with_plus_exponent).parse.should == {matrix: @matrix}
+        StringParser.new(@input_with_minus_exponent).parse.should == {matrix: @matrix}
+        StringParser.new(@input_with_upcase_exponent).parse.should == {matrix: @matrix}
+        StringParser.new(@input_with_manydigit_exponent).parse.should == {matrix: @matrix}
       end
       it 'should raise an error for invalid input string' do
         expect{ StringParser.new(@bad_input_not_numeric).parse }.to raise_error ArgumentError
         expect{ StringParser.new(@bad_input_different_row_size).parse }.to raise_error ArgumentError
         expect{ StringParser.new(@bad_input_not_4_rows_and_cols).parse }.to raise_error ArgumentError
+        expect{ StringParser.new(@bad_input_with_empty_exponent).parse }.to raise_error ArgumentError
       end
     end
   end

data/spec/data_models/pm_spec.rb

@@ -3,7 +3,7 @@ require 'bioinform/data_models/pm'
 
 module Bioinform
   describe PM do 
-    include PM::Parser::Helpers
+    include Parser::Helpers
     
     describe '#valid?' do
       it 'should be true iff an argument is an array of arrays of 4 numerics in a column' do

data/spec/data_models/pwm_spec.rb

@@ -21,5 +21,25 @@ module Bioinform
     
     describe '#gauss_estimation' do
     end
+
+    describe '#score' do
+      let(:pwm) do
+        pwm = PWM.new
+        pwm.matrix = [[10000,20000,30000,40000],[1000,2000,3000,4000],[100,200,300,400],[10,20,30,40],[1,2,3,4]]
+        pwm
+      end
+      it 'should evaluate to score of given word' do
+        pwm.score('aAAAA').should == 11111
+        pwm.score('agata').should == 13141
+        pwm.score('CCGCT').should == 22324
+      end
+      it 'should raise an ArgumentError if word contain bad letter' do
+        expect{ pwm.score('AAAAV') }.to raise_error(ArgumentError)
+      end
+      it 'should raise an ArgumentError if word has size different than size of matrix' do
+        expect{ pwm.score('AAA') }.to raise_error(ArgumentError)
+      end
+    end
+
   end
 end

data/spec/spec_helper.rb

@@ -3,27 +3,25 @@ $LOAD_PATH.unshift File.dirname(__FILE__)
 
 require 'rspec'
 module Bioinform
-  class PM
-    class Parser
-      module Helpers
-        def parser_stub(class_name, can_parse, result)
-          klass = Class.new(PM::Parser) do
-            define_method :can_parse? do can_parse end
-            define_method :parse do result end
-          end
-          #class_levels = class_name.to_s.split('::')
-          #class_levels[0..-2].inject(Object){|klass, level| klass.const_get level}.const_set(class_name, class_levels.last)
-          Bioinform.const_set(class_name.to_s.split('::').last, klass)
+  class Parser
+    module Helpers
+      def parser_stub(class_name, can_parse, result)
+        klass = Class.new(Parser) do
+          define_method :can_parse? do can_parse end
+          define_method :parse do result end
         end
-        def parser_subclasses_cleanup          
-          PM::Parser.subclasses.each do |klass|
-            #class_levels = klass.to_s.split('::')
-            #class_levels[0..-2].inject(Object){|klass, level| klass.const_get level}.const_set(class_name, class_levels.last)
-            
-            Bioinform.send :remove_const, klass.name.split('::').last
-          end
-          PM::Parser.subclasses.clear
+        #class_levels = class_name.to_s.split('::')
+        #class_levels[0..-2].inject(Object){|klass, level| klass.const_get level}.const_set(class_name, class_levels.last)
+        Bioinform.const_set(class_name.to_s.split('::').last, klass)
+      end
+      def parser_subclasses_cleanup          
+        Parser.subclasses.each do |klass|
+          #class_levels = klass.to_s.split('::')
+          #class_levels[0..-2].inject(Object){|klass, level| klass.const_get level}.const_set(class_name, class_levels.last)
+          
+          Bioinform.send :remove_const, klass.name.split('::').last
         end
+        Parser.subclasses.clear
       end
     end
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bioinform
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.1.2
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-06-27 00:00:00.000000000 Z
+date: 2012-07-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -78,7 +78,6 @@ files:
 - lib/bioinform/support.rb
 - lib/bioinform/support/callable_symbol.rb
 - lib/bioinform/support/collect_hash.rb
-- lib/bioinform/support/curry_except_self.rb
 - lib/bioinform/support/deep_dup.rb
 - lib/bioinform/support/delete_many.rb
 - lib/bioinform/support/has_keys.rb
@@ -98,7 +97,6 @@ files:
 - spec/spec_helper.rb
 - spec/support/callable_symbol_spec.rb
 - spec/support/collect_hash_spec.rb
-- spec/support/curry_except_self_spec.rb
 - spec/support/delete_many_spec.rb
 - spec/support/has_keys_spec.rb
 - spec/support/inverf_spec.rb
@@ -143,7 +141,6 @@ test_files:
 - spec/spec_helper.rb
 - spec/support/callable_symbol_spec.rb
 - spec/support/collect_hash_spec.rb
-- spec/support/curry_except_self_spec.rb
 - spec/support/delete_many_spec.rb
 - spec/support/has_keys_spec.rb
 - spec/support/inverf_spec.rb

data/lib/bioinform/support/curry_except_self.rb

@@ -1,5 +0,0 @@
-class Proc
-  def curry_except_self(*args, &block)
-    Proc.new{|slf| curry.call(slf, *args, &block) }
-  end
-end

data/spec/support/curry_except_self_spec.rb

@@ -1,9 +0,0 @@
-require 'spec_helper'
-require 'bioinform/support/curry_except_self'
-
-describe Proc do
-  describe '#curry_except_self' do
-    it 'should return proc'
-    it 'should behave like a proc where all arguments except first are curried'
-  end
-end