rbbt-dm 0.0.3 → 0.0.4

data/lib/rbbt/statistics/hypergeometric.rb

@@ -97,10 +97,13 @@ module TSV
     fields ||= self.fields
     fields = [fields] if String === fields or Symbol === fields
 
-    Persist.persist(filename, :marshal, :fields => fields, :persist => persistence, :prefix => "Hyp.Geo.Counts") do 
+    Persist.persist(filename, :yaml, :fields => fields, :persist => persistence, :prefix => "Hyp.Geo.Counts") do 
       data ||= Hash.new(0)
-      through :key, fields do |key, values|
-        values.flatten.compact.uniq.each{|value| data[value] += 1}
+
+      with_unnamed do
+        through :key, fields do |key, values|
+          values.flatten.compact.uniq.each{|value| data[value] += 1}
+        end
       end
 
       data
@@ -108,40 +111,70 @@ module TSV
   end
 
   def enrichment(list, fields = nil, options = {})
-    fields ||= self.fields.first
-    options = Misc.add_defaults options, :min_support => 3, :fdr => true, :cutoff => false
-    Log.debug "Enrichment analysis of field #{fields.inspect} for #{list.length} entities"
+    with_unnamed do
+      fields ||= self.fields.first
+      options = Misc.add_defaults options, :min_support => 3, :fdr => true, :cutoff => false, :add_keys => true
+
+      add_keys = Misc.process_options options, :add_keys
+
+      Log.debug "Enrichment analysis of field #{fields.inspect} for #{list.length} entities"
+
+      selected = select :key => list
+
+      tsv_size = keys.length
+      total = selected.keys.length
+      Log.debug "Found #{total} of #{list.length} entities"
+
+      counts = annotation_counts fields, options[:persist]
+
+      annotations = Hash.new 
+      annotation_keys = Hash.new
+      selected.with_unnamed do
+        selected.through :key, fields do |key, values|
+          values.flatten.compact.uniq.reject{|value| value.empty?}.each{|value| 
+            value = value.dup
+            annotations[value] ||= 0
+            annotations[value] += 1
+            next unless add_keys
+            annotation_keys[value] ||= []
+            annotation_keys[value] << key
+          }
+        end
+      end
 
-    selected = select :key => list
+      pvalues = {}
+      annotations.each do |annotation, count|
+        next if count < options[:min_support] or not counts.include? annotation
+        pvalues[annotation] = Hypergeometric.hypergeometric(tsv_size, counts[annotation], total, count)
+      end
 
-    tsv_size = keys.length
-    total = selected.keys.length
-    Log.debug "Found #{total} of #{list.length} entities"
+      FDR.adjust_hash! pvalues if options[:fdr]
 
-    counts = annotation_counts fields, options[:persist]
+      pvalues.delete_if{|k, pvalue| pvalue > options[:cutoff] } if options[:cutoff]
 
-    annotations = Hash.new 0
-    with_unnamed do
-      selected.through :key, fields do |key, values|
-        values.flatten.compact.uniq.reject{|value| value.empty?}.each{|value| 
-          annotations[value] += 1
-        }
-      end
-    end
+      TSV.setup(pvalues, :key_field => fields, :fields => ["p-value"], :cast => :to_f, :type => :single)
 
-    pvalues = {}
-    annotations.each do |annotation, count|
-      next if count < options[:min_support] or not counts.include? annotation
-      pvalue = Hypergeometric.hypergeometric(tsv_size, counts[annotation], total, count)
-      pvalues[annotation] = pvalue
-    end
+      if add_keys
+        tsv = TSV.setup(pvalues.keys, :key_field => fields, :fields => [], :type => :double)
 
-    FDR.adjust_hash! pvalues if options[:fdr]
-    pvalues.delete_if{|k, pvalue| pvalue > options[:cutoff] } if options[:cutoff]
+        tsv.add_field 'p-value' do |annot, values|
+          [pvalues[annot]]
+        end
 
-    TSV.setup(pvalues, :key_field => fields, :fields => ["p-value"], :cast => :to_f, :type => :single)
+        tsv.add_field self.key_field do |annot, values|
+          if list.respond_to? :annotate
+            list.annotate annotation_keys[annot]
+          else
+            annotation_keys[annot]
+          end
+        end
 
-    pvalues
+        tsv
+      else
+        pvalues
+      end
+
+    end
   end
 
   def enrichment_for(tsv, field, options = {} )

data/lib/rbbt/statistics/random_walk.rb

@@ -1,5 +1,6 @@
 require 'png'
 require 'inline'
+require 'set'
 
 module RandomWalk
 
@@ -89,8 +90,10 @@ module RandomWalk
     if size == 0
       [0] * times
     else
+      a = (0..total - 1).to_a
       (1..times).collect do
-        score(Array.new(size){ (rand * total).to_i }.sort, total, missing).abs
+        a.shuffle!
+        score(a[1..size].sort, total, missing).abs
       end
     end
   end
@@ -114,7 +117,7 @@ module RandomWalk
     :green => PNG::Color::Green,
     :white => PNG::Color::White,
     :black => PNG::Color::Black,
-
+    :gray => PNG::Color::Gray,
   }
 
   def self.draw_hits(hits, total, filename = nil, options = {})
@@ -131,7 +134,7 @@ module RandomWalk
       hits = hits.collect{|h| (h.to_f * size / total).to_i}
     end
 
-    canvas = PNG::Canvas.new size, width, COLORS[bg_color]
+    canvas = PNG::Canvas.new size, width, COLORS[bg_color] || PNG::Color.from(bg_color)
 
     sections.each{|color, info|
       start = info[0]
@@ -156,3 +159,35 @@ module RandomWalk
     end
   end
 end
+
+module OrderedList
+  def self.hits(list, set)
+    set = Set.new(set) unless Set === set
+    hits = []
+    list.each_with_index do |e,i|
+      hits << i if set.include? e
+    end
+    hits
+  end
+
+  def self.draw_hits(list, set, filename = nil, options = {})
+    hits = OrderedList.hits(list, set)
+    RandomWalk.draw_hits(hits, list.length, filename, options)
+  end
+
+  def hits(set)
+    OrderedList.hits(self, set)
+  end
+
+  def draw_hits(set, filename = nil, options = {})
+    OrderedList.draw_hits(self, set, filename, options)
+  end
+
+  def pvalue(set, options = {})
+    options = Misc.add_defaults options, :permutations => 1000, :missing => 0
+    hits = hits(set.compact)
+    score = RandomWalk.score(hits.sort, self.length, 0)
+    permutations = RandomWalk.permutations(set.length, self.length, options[:missing], options[:permutations])
+    RandomWalk.pvalue(permutations, score)
+  end
+end

data/lib/rbbt/statistics/rank_product.rb

@@ -0,0 +1,68 @@
+require 'rbbt/tsv'
+
+module RankProduct
+  def self.score(gene_ranks, signature_sizes)
+    scores = {}
+    log_sizes = signature_sizes.collect{|size| Math::log(size)}
+    gene_ranks.each{|gene, positions|
+      scores[gene] = positions.zip(log_sizes).
+        collect{|p| Math::log(p[0]) - p[1]}.    # Take log and substract from size (normalize)
+        inject(0){|acc, v| acc += v  }
+    }
+    scores
+  end
+
+  def self.permutations(num_signatures, num = 1000)
+    scores = []
+    num.times{
+       value = 0
+       num_signatures.times{|size_and_log| 
+         value += Math::log(rand)
+       } 
+       scores << value
+    }
+    scores
+  end
+
+  def self.permutations_full(signature_sizes)
+    gene_ranks = {}
+    signature_sizes.each{|size|
+      (1..size).to_a.shuffle.each_with_index{|gene, pos|
+        gene_ranks[gene] ||= []
+        gene_ranks[gene] << pos + 1
+      }
+    }
+    gene_ranks.delete_if{|code, positions| positions.length != signature_sizes.length}
+
+    scores = score(gene_ranks, signature_sizes)
+    scores.values
+  end
+end
+
+module TSV
+  def rank_product(fields, reverse = false, &block)
+    tsv = self.slice(fields)
+
+    if block_given?
+      scores = fields.collect{|field| tsv.sort_by(field, true, &block)}
+    else
+      scores = fields.collect{|field| tsv.sort_by(field, true){|gene,values| tsv.type == :double ? values.first.to_f : value.to_f}}
+    end
+    positions = {}
+    
+    if reverse
+      size = self.size
+      tsv.keys.each do |entity|
+        positions[entity] = scores.collect{|list| size - list.index(entity)}
+      end
+    else
+      tsv.keys.each do |entity|
+        positions[entity] = scores.collect{|list| list.index(entity) + 1}
+      end
+    end
+
+    score = RankProduct.score(positions, fields.collect{ tsv.size })
+
+    score
+  end
+end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: rbbt-dm
 version: !ruby/object:Gem::Version 
-  hash: 25
+  hash: 23
   prerelease: 
   segments: 
   - 0
   - 0
-  - 3
-  version: 0.0.3
+  - 4
+  version: 0.0.4
 platform: ruby
 authors: 
 - Miguel Vazquez
@@ -15,8 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-10-04 00:00:00 +02:00
-default_executable: 
+date: 2012-01-13 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: rbbt-util
@@ -33,7 +32,7 @@ dependencies:
   type: :runtime
   version_requirements: *id001
 - !ruby/object:Gem::Dependency 
-  name: rbbt-entities
+  name: RubyInline
   prerelease: false
   requirement: &id002 !ruby/object:Gem::Requirement 
     none: false
@@ -47,7 +46,7 @@ dependencies:
   type: :runtime
   version_requirements: *id002
 - !ruby/object:Gem::Dependency 
-  name: RubyInline
+  name: priority_queue
   prerelease: false
   requirement: &id003 !ruby/object:Gem::Requirement 
     none: false
@@ -60,20 +59,6 @@ dependencies:
         version: "0"
   type: :runtime
   version_requirements: *id003
-- !ruby/object:Gem::Dependency 
-  name: priority_queue
-  prerelease: false
-  requirement: &id004 !ruby/object:Gem::Requirement 
-    none: false
-    requirements: 
-    - - ">="
-      - !ruby/object:Gem::Version 
-        hash: 3
-        segments: 
-        - 0
-        version: "0"
-  type: :runtime
-  version_requirements: *id004
 description: Data-mining and statistics
 email: miguel.vazquez@fdi.ucm.es
 executables: []
@@ -88,6 +73,7 @@ files:
 - lib/rbbt/statistics/fdr.rb
 - lib/rbbt/statistics/hypergeometric.rb
 - lib/rbbt/statistics/random_walk.rb
+- lib/rbbt/statistics/rank_product.rb
 - lib/rbbt/vector/model.rb
 - lib/rbbt/vector/model/svm.rb
 - test/rbbt/statistics/test_fdr.rb
@@ -96,7 +82,6 @@ files:
 - test/rbbt/vector/model/test_svm.rb
 - test/rbbt/network/test_paths.rb
 - test/test_helper.rb
-has_rdoc: true
 homepage: http://github.com/mikisvaz/rbbt-phgx
 licenses: []
 
@@ -126,7 +111,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.6.2
+rubygems_version: 1.8.10
 signing_key: 
 specification_version: 3
 summary: Data-mining and statistics