NetAnalyzer 0.1.5 → 0.6.3

data/lib/NetAnalyzer/nodes.rb

@@ -4,4 +4,19 @@ class Node
 		@id = id
 		@type = type
 	end
+
+	def clone
+		node_clone = Node.new(@id.clone, @type.clone)
+		return node_clone
+	end
+
+	def ==(other)
+		are_equal = true
+		if self.id != other.id ||
+		   self.type != other.type	
+		   are_equal = false
+		end   
+		return are_equal
+	end
+
 end

data/lib/NetAnalyzer/performancer.rb

@@ -0,0 +1,98 @@
+require 'bigdecimal'
+
+class Performancer
+	def initialize()
+		@control = {}
+	end
+
+	def load_control(ref_array)
+		ref_array.each do |node1, node2|
+			if node2 != '-'
+				query = @control[node1]
+				if query.nil?
+					@control[node1] = [node2]
+				else
+					query << node2
+				end
+			end
+		end
+		return control
+	end
+
+	# Pandey 2007, Association Analysis-based Transformations for Protein Interaction Networks: A Function Prediction Case Study
+	def get_pred_rec(predictions, cut_number = 100, top_number = 10000)
+		performance = [] #cut, pred, rec
+		preds, limits = load_prediction(predictions)
+		cuts = get_cuts(limits, cut_number)
+		cuts.each do |cut|
+			prec, rec = pred_rec(preds, cut, top_number)
+			performance << [cut, prec, rec]
+		end
+		return performance
+	end
+
+	def load_prediction(pairs_array)
+		pred = {}
+		min = nil
+		max = nil
+		pairs_array.each do |key, label, score|
+			query = pred[key]
+			if !min.nil? && !max.nil?
+				min = score if score < min
+				max = score if score > max
+			else
+				min = score; max = score
+			end
+			if query.nil?
+				pred[key] = [[label], [score]]
+			else
+				query.first << label
+				query.last << score
+			end
+		end
+		return pred, [min, max]
+	end
+
+	def pred_rec(preds, cut, top)
+		predicted_labels = 0 #m
+		true_labels = 0 #n
+		common_labels = 0 # k
+		@control.each do |key, c_labels|
+			true_labels += c_labels.length #n
+			pred_info = preds[key]
+			if !pred_info.nil?
+				labels, scores = pred_info
+				reliable_labels = get_reliable_labels(labels, scores, cut, top)
+				predicted_labels += reliable_labels.length #m
+				common_labels += (c_labels & reliable_labels).length #k
+			end
+		end
+		#puts "cut: #{cut} trueL: #{true_labels} predL: #{predicted_labels} commL: #{common_labels}"
+		prec = common_labels.to_f/predicted_labels
+		rec = common_labels.to_f/true_labels
+		prec = 0.0 if prec.nan?
+		rec = 0.0 if rec.nan?
+		return prec, rec
+	end
+
+
+	def get_cuts(limits, n_cuts)
+		cuts = []
+		range = (limits.last - limits.first).abs.fdiv(n_cuts)
+		range = BigDecimal(range, 10)
+		cut = limits.first
+		(n_cuts + 1).times do |n|
+			cuts << (cut + n * range).to_f
+		end
+		return cuts
+	end
+
+	def get_reliable_labels(labels, scores, cut, top)
+		reliable_labels = []
+		scores.each_with_index do |score, i|
+			reliable_labels << [labels[i], score] if score >= cut
+		end
+		reliable_labels = reliable_labels.sort!{|l1,l2| l2.last <=> l1.last}[0..top-1].map{|pred| pred.first}
+		return reliable_labels
+	end
+end

data/lib/NetAnalyzer/ranker.rb

@@ -0,0 +1,250 @@
+#!/usr/bin/env ruby
+
+require 'numo/narray'
+require "numo/linalg"
+
+class Ranker
+  attr_accessor :matrix, :nodes, :seeds, :ranking
+
+  def initialize()
+    @matrix = nil
+    @nodes = [] # kernel_nodes
+    @seeds = {} # genes_seed
+    @reference_nodes = {}
+    @ranking = {} # ranked_genes
+  end
+
+  def load_seeds(node_groups, sep: ',')
+    @seeds = load_nodes_by_group(node_groups, sep: sep)
+  end
+
+  def load_references(node_groups, sep: ',')
+    @reference_nodes = load_nodes_by_group(node_groups, sep: sep)
+  end
+
+  def load_nodes_by_group(node_groups, sep: ',')
+    if File.exist?(node_groups) 
+      group_nodes = load_node_groups_from_file(node_groups, sep: sep)
+    else
+      group_nodes = {"seed_genes" => node_groups.split(sep)}
+    end
+    return group_nodes
+  end
+          
+  def load_node_groups_from_file(file, sep: ',')
+    group_nodes = {}
+    File.open(file).each do |line|
+      set_name, nodes = line.chomp.split("\t")
+      group_nodes[set_name] = nodes.split(sep)
+    end
+    return group_nodes
+  end
+
+  def load_nodes_from_file(file)
+    File.open(file).each do |line|
+      @nodes << line.chomp!
+    end
+  end
+
+  def do_ranking(leave_one_out: false, threads: 0)
+    get_seed_leave_one_out() if leave_one_out
+    seed_indexes = get_seed_indexes
+    seed_groups = @seeds.to_a # Array conversion needed for parallelization
+    ranked_lists = Parallel.map(seed_groups, in_processes: threads) do |seed_name, seed|
+      # The code in this block CANNOT modify nothing outside
+      if leave_one_out and @reference_nodes[seed_name].length <= 1
+         rank_list = get_individual_rank(seed,@reference_nodes[seed_name][0])
+      else
+         rank_list = rank_by_seed(seed_indexes, seed) # Production mode
+      end
+      [seed_name, rank_list] 
+    end
+    ranked_lists.each do |seed_name, rank_list| # Transfer resuls to hash
+      @ranking[seed_name] = rank_list
+    end
+  end
+
+  def get_seed_leave_one_out()
+    new_seeds = {}
+    genes2predict = {}
+    all_genes = @seeds.values.flatten.uniq
+    @seeds.each do |seed_name, seeds| 
+      group_number = seeds.length - 1
+      one_out_seeds = seeds.combination(group_number).to_a
+
+      one_out_seeds.each_with_index do |one_out_seed, indx|
+        seed_name_one_out = seed_name.to_s + "_iteration_" + indx.to_s
+        new_seeds[seed_name_one_out] = one_out_seed
+        genes2predict[seed_name_one_out] = seeds - one_out_seed
+        genes2predict[seed_name_one_out] += @reference_nodes[seed_name] if !@reference_nodes[seed_name].nil?
+        genes2predict[seed_name_one_out].uniq!
+      end
+    end
+    @seeds = new_seeds
+    @reference_nodes = genes2predict
+  end
+
+  def rank_by_seed(seed_indexes, seeds)
+    ordered_gene_score = []
+    genes_pos = seeds.map{|s| seed_indexes[s]}.compact
+    number_of_seed_genes = genes_pos.length
+    number_of_all_nodes = @nodes.length
+    
+    if number_of_seed_genes > 0
+      subsets_gen_values = @matrix[genes_pos,true]
+      integrated_gen_values = subsets_gen_values.sum(0)
+      gen_list = 1.fdiv(number_of_seed_genes) * integrated_gen_values.inplace
+
+      ordered_indexes = gen_list.sort_index # from smallest to largest
+      
+      last_val = nil
+      n_elements = ordered_indexes.shape.first 
+      n_elements.times do |pos|
+        order_index = ordered_indexes[pos]    
+        val = gen_list[order_index]
+        node_name = @nodes[order_index]
+
+        rank = get_position_for_items_with_same_score(pos, val, last_val, gen_list, n_elements, ordered_gene_score) # number of items behind
+        rank = n_elements - rank # number of nodes below or equal
+        rank_percentage = rank.fdiv(number_of_all_nodes)
+
+        ordered_gene_score << [node_name, val, rank_percentage, rank]
+        last_val = val
+      end
+
+      ordered_gene_score = ordered_gene_score.reverse # from largest to smallest
+      ordered_gene_score = add_absolute_rank_column(ordered_gene_score)
+    end
+    
+    return ordered_gene_score
+  end
+
+  def get_position_for_items_with_same_score(pos, val, prev_val, gen_list, n_elements, ordered_gene_score)
+      members_behind = 0
+      if !prev_val.nil?
+        if prev_val < val
+          members_behind = pos
+        else 
+          members_behind = n_elements - ordered_gene_score.last[3]
+        end
+      end
+      return members_behind
+  end
+
+  def add_absolute_rank_column(ranking)
+    ranking_with_new_column = []
+    absolute_rank = 1
+    n_rows = ranking.length
+    n_rows.times do |row_pos|
+      if row_pos == 0
+        new_row = ranking[row_pos] << absolute_rank
+        ranking_with_new_column << new_row
+      else
+        prev_val = ranking[row_pos-1][2]
+        val = ranking[row_pos][2]
+        if val > prev_val
+          absolute_rank +=1
+        end
+        new_row = ranking[row_pos] << absolute_rank
+        ranking_with_new_column << new_row
+      end
+    end
+    return ranking_with_new_column
+  end
+
+  def get_individual_rank(seed_genes, node_of_interest)
+    genes_pos = get_nodes_indexes(seed_genes)
+    node_of_interest_pos = @nodes.find_index(node_of_interest)
+    ordered_gene_score = []
+
+    if !genes_pos.empty? && !node_of_interest_pos.nil?
+
+      subsets_gen_values = @matrix[genes_pos,true]
+      integrated_gen_values = subsets_gen_values.sum(0)
+      integrated_gen_values = 1.fdiv(genes_pos.length) * integrated_gen_values.inplace
+
+      ref_value = integrated_gen_values[node_of_interest_pos]
+
+      members_below_test = 0
+      integrated_gen_values.each do |gen_value|
+        members_below_test += 1 if gen_value >= ref_value
+      end
+
+      rank_percentage = members_below_test.fdiv(@nodes.length)
+      rank = members_below_test
+      rank_absolute = get_individual_absolute_rank(integrated_gen_values.to_a,ref_value)
+
+      ordered_gene_score << [node_of_interest, ref_value, rank_percentage, rank, rank_absolute]
+    end
+
+    return ordered_gene_score
+
+  end
+
+  def get_individual_absolute_rank(values_list,ref_value)
+    ref_pos = nil
+    values_list = values_list.sort.reverse.uniq
+    values_list.each_with_index do |value,pos|
+      if value == ref_value
+        ref_pos = pos+1
+        break
+      end
+    end
+    return ref_pos
+  end
+
+  def get_reference_ranks
+    filtered_ranked_genes = {}
+    
+    @ranking.each do |seed_name, ranking|
+      next if @reference_nodes[seed_name].nil? or ranking.empty?
+
+      ranking = array2hash(ranking,0,(1..ranking[0].length))
+      references = @reference_nodes[seed_name]
+      filtered_ranked_genes[seed_name] = []
+
+      references.each do |reference|
+        rank = ranking[reference]
+        if !rank.nil?
+          filtered_ranked_genes[seed_name] << [reference] + rank
+        end
+      end
+      filtered_ranked_genes[seed_name].sort_by!{|rank| -rank[1]}
+    end
+    return filtered_ranked_genes
+  end
+
+  def array2hash(arr, key, values)
+    h = {}
+    arr.each{|els| h[els[0]] = els[values]}
+    return h
+  end
+
+  def get_top(top_n)
+  	top_ranked_genes = {}
+  	@ranking.each do |seed_name, ranking|
+        top_ranked_genes[seed_name] = ranking[0..top_n-1] if !ranking.nil?
+  	end
+  	return top_ranked_genes
+  end
+
+  def get_nodes_indexes(nodes)
+    node_indxs = []
+    nodes.each do |node|
+      index_node = @nodes.find_index(node)
+      node_indxs << index_node if !index_node.nil?
+    end
+    return node_indxs
+  end
+
+  def get_seed_indexes
+    indexes = {}
+    @seeds.values.flatten.each do |node|
+      if !indexes.include?(node)
+        indx = @nodes.index(node)
+        indexes[node] = indx if !indx.nil?
+      end
+    end
+    return indexes
+  end
+end