bioinform 0.1.2 → 0.1.3

data/lib/bioinform/data_models/pm.rb

@@ -104,37 +104,31 @@ module Bioinform
     end
     
     def background!(new_background)
-      clear_cache
       @background = new_background
       self
     end
     
     def self.zero_column
-      [0.0, 0.0, 0.0, 0.0]
+      [0, 0, 0, 0]
     end
 
-    def reverse_complement!
-      clear_cache
+    def reverse_complement!      
       @matrix.reverse!.map!(&:reverse!)
       self
     end
     def left_augment!(n)
-      clear_cache
       n.times{ @matrix.unshift(self.class.zero_column) }
       self
     end
     def right_augment!(n)
-      clear_cache
       n.times{ @matrix.push(self.class.zero_column) }
       self
     end
     def shift_to_zero! # make worst score == 0 by shifting scores of each column
-      clear_cache
       @matrix.map!{|position| min = position.min; position.map{|element| element - min}}
       self
     end
     def discrete!(rate)
-      clear_cache
       @matrix.map!{|position| position.map{|element| (element * rate).ceil}}
       self
     end
@@ -161,23 +155,19 @@ module Bioinform
     #end
 
     def best_score
-      @best_score ||= @matrix.inject(0.0){|sum, col| sum + col.max}
+      @matrix.inject(0.0){|sum, col| sum + col.max}
     end
     def worst_score
-      @worst_score ||= @matrix.inject(0.0){|sum, col| sum + col.min}
+      @matrix.inject(0.0){|sum, col| sum + col.min}
     end
     
     # best score of suffix s[i..l]
-    def best_suffix
-      @best_suffix ||= Array.new(length + 1) {|i| @matrix[i...length].map(&:max).inject(0.0, &:+) }
+    def best_suffix(i)
+      @matrix[i...length].map(&:max).inject(0.0, &:+)
     end
     
-    def worst_suffix
-      @worst_suffix ||= Array.new(length + 1) {|i| @matrix[i...length].map(&:min).inject(0.0, &:+) }
-    end
-    
-    def clear_cache
-      @best_score, @worst_score, @best_suffix, @worst_suffix = nil,nil,nil,nil
+    def worst_suffix(i)
+      @matrix[i...length].map(&:min).inject(0.0, &:+)
     end
     
     def reverse_complement

data/lib/bioinform/support.rb

@@ -11,4 +11,7 @@ require 'bioinform/support/yaml_dump_file'
 require 'bioinform/support/inverf'
 require 'bioinform/support/deep_dup'
 
-require 'bioinform/support/partial_sums'
+require 'bioinform/support/partial_sums'
+
+require 'bioinform/support/array_zip'
+require 'bioinform/support/array_product'

data/lib/bioinform/support/array_product.rb

@@ -0,0 +1,6 @@
+class Array
+  def self.product(*arrays)
+    return []  if arrays.empty?
+    arrays.first.product(*arrays[1..-1])
+  end
+end

data/lib/bioinform/support/array_zip.rb

@@ -0,0 +1,6 @@
+class Array
+  def self.zip(*arrays)
+    return []  if arrays.empty?
+    arrays.first.zip(*arrays[1..-1])
+  end
+end

data/lib/bioinform/support/partial_sums.rb

@@ -4,3 +4,11 @@ class Array
     map{|el| sums += el}
   end
 end
+
+class Hash
+# {1 => 5, 4 => 3, 3 => 2}.partial_sums == {1=>5, 3=>7, 4=>10}
+  def partial_sums(initial = 0.0)
+    sums = initial
+    sort.collect_hash{|k,v| [k, sums += v]}
+  end
+end

data/lib/bioinform/version.rb

@@ -1,3 +1,3 @@
 module Bioinform
-  VERSION = "0.1.2"
+  VERSION = "0.1.3"
 end

data/spec/data_models/pm_spec.rb

@@ -318,17 +318,23 @@ module Bioinform
     end
     
     describe '#best_suffix' do
-      it 'should be an array of best suffices from start of string and to empty suffix inclusive' do
+      it 'should return maximal score of suffices from i-th position inclusively i.e. [i..end]' do
         @pm = PM.new
         @pm.matrix = [[1.3, 2.0, 4.9, 3.2], [7.13, 6.5, 3.25, 4.633], [-1.0, -1.0, -1.5, -1.0]]
-        @pm.best_suffix.should == [(4.9 + 7.13 - 1.0), (7.13 - 1.0), (-1.0), (0.0) ]
+        @pm.best_suffix(0).should == (4.9 + 7.13 - 1.0)
+        @pm.best_suffix(1).should == (7.13 - 1.0)
+        @pm.best_suffix(2).should == (-1.0)
+        @pm.best_suffix(3).should == (0.0)
       end
     end
     describe '#worst_suffix' do
-      it 'should be an array of worst suffices from start of string and to empty suffix inclusive' do
+      it 'should return minimal score of suffices from i-th position inclusively i.e. [i..end]' do
         @pm = PM.new
         @pm.matrix = [[1.3, 2.0, 4.9, 3.2], [7.13, 6.5, 3.25, 4.633], [-1.0, -1.0, -1.5, -1.0]]
-        @pm.worst_suffix.should == [(1.3 + 3.25 - 1.5), (3.25 - 1.5), (- 1.5), (0.0) ]
+        @pm.worst_suffix(0).should == (1.3 + 3.25 - 1.5)
+        @pm.worst_suffix(1).should == (3.25 - 1.5)
+        @pm.worst_suffix(2).should == (- 1.5)
+        @pm.worst_suffix(3).should == (0.0)
       end
     end
 

data/spec/support/array_product_spec.rb

@@ -0,0 +1,15 @@
+require 'spec_helper'
+require 'bioinform/support/array_product'
+
+describe Array do
+  context '::product' do
+    it 'should take any number of arrays and product them as if #product was made to first and others' do
+      Array.product([1,2,3]).should == [1,2,3].product()
+      Array.product([1,2,3],[4,5,6]).should == [1,2,3].product([4,5,6])
+      Array.product([1,2,3],[4,5,6],[7,8,9]).should == [1,2,3].product([4,5,6],[7,8,9])
+    end
+    it 'should return empty array if no arrays\'re given' do
+      Array.product().should == []
+    end
+  end
+end

data/spec/support/array_zip_spec.rb

@@ -0,0 +1,15 @@
+require 'spec_helper'
+require 'bioinform/support/array_zip'
+
+describe Array do
+  context '::zip' do
+    it 'should take any number of arrays and zip them as if #zip was made to first and others' do
+      Array.zip([1,2,3]).should == [1,2,3].zip()
+      Array.zip([1,2,3],[4,5,6]).should == [1,2,3].zip([4,5,6])
+      Array.zip([1,2,3],[4,5,6],[7,8,9]).should == [1,2,3].zip([4,5,6],[7,8,9])
+    end
+    it 'should return empty array if no arrays\'re given' do
+      Array.zip().should == []
+    end
+  end
+end

data/spec/support/partial_sums_spec.rb

@@ -1,9 +1,27 @@
 describe 'Array#partial_sums' do
-  it 'should return an array of the same size with partial sums of elements 0..ind inclusive with float elements' do
-    [2,3,4,5].partial_sums.should == [2, 5, 9, 14]
-    [2,3,4,5].partial_sums.last.should be_kind_of(Float)
+  context 'when no initial value given' do
+    it 'should return an array of the same size with partial sums of elements 0..ind inclusive with float elements' do
+      [2,3,4,5].partial_sums.should == [2, 5, 9, 14]
+      [2,3,4,5].partial_sums.last.should be_kind_of(Float)
+    end
   end
-  it 'should start counting from argument when it\'s given' do
+  it 'should start counting from argument when it\'s given. Type of values depends on type of initial value' do
     [2,3,4,5].partial_sums(100).should == [102,105,109,114]
+    [2,3,4,5].partial_sums(100).last.should be_kind_of(Integer)
   end
-end
+end
+
+{1 => 5, 4 => 3, 3 => 2}.partial_sums == {1=>5, 3=>7, 4=>10}
+
+describe 'Hash#partial_sums' do
+  context 'when no initial value given' do
+    it 'should return a hash with float values of the same size with partial sums of elements that has keys <= than argument' do
+      {1 => 5, 4 => 3, 3 => 2}.partial_sums.should == {1=>5, 3=>7, 4=>10}
+      {1 => 5, 4 => 3, 3 => 2}.partial_sums.values.last.should be_kind_of(Float)
+    end
+  end
+  it 'should start counting from argument when it\'s given. Type of values depends on type of initial value' do
+    {1 => 5, 4 => 3, 3 => 2}.partial_sums(100).should == {1=>105, 3=>107, 4=>110}
+    {1 => 5, 4 => 3, 3 => 2}.partial_sums(100).values.last.should be_kind_of(Integer)
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bioinform
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.1.3
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-07-07 00:00:00.000000000 Z
+date: 2012-07-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -59,12 +59,6 @@ files:
 - lib/bioinform.rb
 - lib/bioinform/data_models.rb
 - lib/bioinform/data_models/collection.rb
-- lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/iupac.rb
-- lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/iupac_word.rb
-- lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_count_matrix.rb
-- lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_matrix.rb
-- lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_probability_matrix.rb
-- lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_weight_matrix.rb
 - lib/bioinform/data_models/parser.rb
 - lib/bioinform/data_models/parsers.rb
 - lib/bioinform/data_models/parsers/array_parser.rb
@@ -76,6 +70,8 @@ files:
 - lib/bioinform/data_models/ppm.rb
 - lib/bioinform/data_models/pwm.rb
 - lib/bioinform/support.rb
+- lib/bioinform/support/array_product.rb
+- lib/bioinform/support/array_zip.rb
 - lib/bioinform/support/callable_symbol.rb
 - lib/bioinform/support/collect_hash.rb
 - lib/bioinform/support/deep_dup.rb
@@ -95,6 +91,8 @@ files:
 - spec/data_models/pm_spec.rb
 - spec/data_models/pwm_spec.rb
 - spec/spec_helper.rb
+- spec/support/array_product_spec.rb
+- spec/support/array_zip_spec.rb
 - spec/support/callable_symbol_spec.rb
 - spec/support/collect_hash_spec.rb
 - spec/support/delete_many_spec.rb
@@ -139,6 +137,8 @@ test_files:
 - spec/data_models/pm_spec.rb
 - spec/data_models/pwm_spec.rb
 - spec/spec_helper.rb
+- spec/support/array_product_spec.rb
+- spec/support/array_zip_spec.rb
 - spec/support/callable_symbol_spec.rb
 - spec/support/collect_hash_spec.rb
 - spec/support/delete_many_spec.rb

data/lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/iupac.rb

@@ -1,22 +0,0 @@
-class Iupac
-  attr_reader :words
-  def initialize(input)
-    case input
-    when Array
-      @words = input.map{|word| IupacWord.new word}
-    when String
-      @words = input.gsub("\r\n","\n").split("\n").map{|word| IupacWord.new(word)}
-    when IupacWord
-      @words = [input]
-    else raise ArgumentError, 'Can\'t create IUPAC Word List: unknown input type'
-    end
-    raise ArgumentError, 'IUPAC words should be of the same length' unless @words.same_by?(&:length)
-  end
-  
-  def to_pcm
-    @words.map(&:to_pcm).inject(:+)
-  end
-  def to_pwm
-    to_pcm.to_pwm
-  end
-end

data/lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/iupac_word.rb

@@ -1,27 +0,0 @@
-class IupacWord
-  IupacLetters = %w{A C G T  R Y K M S W  B D H V  N}
-  Code = {"A" => "A", "C" => "C", "G" => "G", "T" => "T", 
-        "AG" => "R", "CT" => "Y", "GT" => "K", "AC" => "M", 
-        "CG" => "S", "AT" => "W", "CGT" => "B", "AGT" => "D", "ACT" => "H", "ACG" => "V", "ACGT" => "N"}
-  Decode = Code.invert
-  LetterCode = Hash[Decode.map{|k,v| 
-    [k, %w{A C G T}.map{|chr| (v.include?(chr) ? 1.0 : 0.0) / v.size} ]
-  }]
-  
-  attr_reader :word
-  attr_accessor :weight
-  def initialize(word)
-    raise "Non-IUPAC letter in a word #{word}" unless word.each_char.all?{|letter| IupacLetters.include? letter}
-    @word = word
-    @weight = 1
-  end
-  
-  def length
-    word.length
-  end
-
-  def to_pcm
-    matrix = @word.each_char.map{|letter| LetterCode[letter]}
-    PositionalCountMatrix.new(matrix)
-  end
-end

data/lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_count_matrix.rb

@@ -1,26 +0,0 @@
-class PositionalCountMatrix < PositionalMatrix
-  attr_accessor :pseudocount, :background
-  def initialize(*args)
-    super
-    raise ArgumentError, 'PCM has negative matrix elements' unless @matrix.all?{|position| position.all?{|el| el>=0 }}
-    raise ArgumentError, 'PCM summary count is zero or negative' unless count>=0
-    # summary counts can slightly differ from each other due to floating point precision
-    unless @matrix.all?{|position| (position.inject(:+) - count).abs < 0.01*count } 
-      raise ArgumentError, 'PCM has different summary count at each position'
-    end
-    @background = [1.0, 1.0, 1.0, 1.0]
-    @pseudocount = 1.0
-  end
-  def count
-    @count ||= @matrix.first.inject(&:+)
-  end
-  def to_pwm
-    normalize_coef = background.inject(&:+)
-    bckgr = @background.map{|el| el*1.0/normalize_coef}
-    PositionalWeightMatrix.new @matrix.map{|pos| pos.map.with_index{|el,ind| Math.log(el+bckgr[ind]*@pseudocount /(bckgr[ind]*(count + @pseudocount))) }}
-  end
-  def +(another_pcm)
-    raise ArgumentError, 'another PCM should be of the same length' unless another_pcm.length == length
-    PositionalCountMatrix.new matrix.map.with_index {|pos, i| pos.map.with_index{|el,j| el+another_pcm.matrix[i][j] }}
-  end
-end

data/lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_matrix.rb

@@ -1,78 +0,0 @@
-require 'active_support/core_ext/hash/indifferent_access'
-require 'bioinform/support'
-
-class PositionalMatrix
-  module DefaultParser
-    number_pat = '[+-]?\d+(\.\d+)?'
-    row_pat = "(#{number_pat} )*#{number_pat}"
-    name_pat = '>? ?(?<name>[\w.-]+)\n'
-    matrix_pat = "(?<matrix>(#{row_pat}\n)*#{row_pat})"
-    Pattern = /\A(#{name_pat})?#{matrix_pat}\z/
-    def self.parse(input)
-      input.multiline_squish.match(Pattern)
-    end
-  end
-  
-  module FantomParser
-    number_pat = '[+-]?\d+(\.\d+)?'
-    row_pat = "(#{number_pat} )*#{number_pat}"
-    matrix_pat = "(?<matrix>(#{row_pat}\n)*#{row_pat})"
-    Pattern = /\ANA (?<name>.+)\nA C G T\n#{matrix_pat}\z/
-    def self.trim_first_position(input)
-      inp = input.split("\n")
-      ([inp[0]] + inp[1..-1].map{|x| x.split(' ')[1..4].join(' ') }).join("\n")
-    end
-    def self.parse(input)
-      trim_first_position(input.multiline_squish).match(Pattern)
-    end
-  end
-  
-  
-  
-  attr_reader :name, :matrix
-  def initialize(input,parser = DefaultParser)
-    case input
-    when String
-      match = parser.parse(input)
-      raise ArgumentError, 'Can\'t create positional matrix basing on such input' unless match
-      @name = match[:name]
-      @matrix = match[:matrix].split("\n").map{|row| row.split.map(&:to_f)}
-    when Hash 
-      inp = input.with_indifferent_access
-      @matrix = [inp[:A],inp[:C], inp[:G], inp[:T]]
-    when Array
-      @matrix = input.map do |pos|
-        case pos
-        when Array then pos
-        when Hash  then [pos[:A], pos[:C], pos[:G], pos[:T]]
-        else raise ArgumentError, 'Unknown type of argument inner dimension'
-        end
-      end
-    else
-      raise ArgumentError, 'Unknown format of input: only Strings, Arrays and hashes\'re available'
-    end
-    raise ArgumentError, 'Input has the different number of columns in each row' unless @matrix.same_by?(&:length)
-    raise ArgumentError unless @matrix.size == 4 || @matrix.first.size == 4
-    @matrix = @matrix.transpose if @matrix.first.size != 4
-  end
-  
-  def size
-    @matrix.size
-  end
-  alias_method :length, :size
-  
-  def to_s(with_name = true)
-    mat_str = @matrix.map(&:join.("\t")).join("\n")
-    (with_name && @name) ? "#{@name}\n#{mat_str}" : mat_str
-  end
-  
-  def pretty_string(with_name = true)
-    header = "   A      C      G      T   \n"
-    mat_str = @matrix.map{|position|  position.map{|el| el.round(3).to_s.rjust(6)}.join(' ') }.join("\n")
-    (with_name && @name)  ?  @name + "\n" + header + mat_str  :  header + mat_str
-  end
-  
-  def to_hash
-    {A: @matrix.map{|pos| pos[0]}, C: @matrix.map{|pos| pos[1]}, G: @matrix.map{|pos| pos[2]}, T: @matrix.map{|pos| pos[3]}}.with_indifferent_access
-  end
-end

data/lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_probability_matrix.rb

@@ -1,17 +0,0 @@
-class PositionalProbabilityMatrix < PositionalMatrix
-  attr_accessor :count
-  def initialize(input_string)
-    super(input_string)
-    raise ArgumentError, 'PPM has negative matrix elements' unless @matrix.all?{|position| position.all?{|el| el>=0 }} 
-    # summary counts can slightly differ from each other due to floating point precision
-    unless @matrix.all?{|position| (position.inject(:+) - 1.0).abs < 0.01 }
-      raise ArgumentError, 'PPM has summary probability at some position that differs from 1.0'
-    end
-  end
-  def to_pcm
-    PositionalCountMatrix.new @matrix.map{|pos| pos.map{|el| el*@count}}
-  end
-  def to_pwm
-    to_pcm.to_pwm
-  end
-end

data/lib/bioinform/data_models/old_style_models_TO_BE_REMOVED/positional_weight_matrix.rb

@@ -1,3 +0,0 @@
-class PositionalWeightMatrix < PositionalMatrix
-
-end