rbbt-study 0.2.0

data/lib/rbbt/entity/study/genotypes/genes.rb

@@ -0,0 +1,103 @@
+module Study
+  property :genes_with_overlapping_mutations => :single do
+    mutations = cohort.metagenotype
+    mutations.genes.compact.flatten.uniq
+  end
+
+  property :altered_isoforms => :single do
+    mutated_isoforms = cohort.metagenotype.subset(relevant_mutations).mutated_isoforms.compact.flatten.uniq
+    return [] if mutated_isoforms.empty?
+    mutated_isoforms.select_by(:consequence){|c| c != "SYNONYMOUS"}
+  end
+
+  property :genes_with_altered_isoform_sequence => :single do
+    altered_isoforms = self.altered_isoforms
+    return [] if altered_isoforms.empty?
+    altered_isoforms.transcript.compact.gene.uniq
+  end
+
+  property :damaged_isoforms => :single do |*args|
+    altered_isoforms = self.altered_isoforms
+    return [] if altered_isoforms.empty?
+    altered_isoforms.select_by(:damaged?, *args)
+  end
+
+  property :genes_with_damaged_isoforms => :single do |*args|
+    damaged_isoforms = damaged_isoforms(*args)
+    return [] if damaged_isoforms.empty?
+    damaged_isoforms.transcript.gene.uniq
+  end
+
+  property :genes_with_affected_splicing_sites => :single do
+    cohort.metagenotype.subset(relevant_mutations).transcripts_with_affected_splicing.compact.flatten.uniq.gene.compact.uniq
+  end
+
+  property :affected_genes => :single do
+    Gene.setup(genes_with_altered_isoform_sequence + genes_with_affected_splicing_sites, "Ensembl Gene ID", organism).uniq
+  end
+
+  property :damaged_genes => :single do |*args|
+    Gene.setup((genes_with_damaged_isoforms(*args) + genes_with_affected_splicing_sites).uniq, "Ensembl Gene ID", organism)
+  end
+
+ property :samples_with_gene_damaged => :single do
+    damaging_mutations= self.damaging_mutations
+
+    samples_with_gene_damaged = {}
+    cohort.each do |genotype|
+      genotype.each do |mutation|
+        next unless damaging_mutations.include? mutation
+        genes = []
+        mis = mutation.mutated_isoforms
+        genes.concat mis.select_by(:damaged?).transcript.gene unless mis.nil? or mis.empty?
+        genes.concat mutation.transcripts_with_affected_splicing.gene
+        genes.uniq.each{|gene| samples_with_gene_damaged[gene] ||= []; samples_with_gene_damaged[gene] << genotype.jobname}
+      end
+    end
+    samples_with_gene_damaged
+  end
+
+  property :samples_with_gene_affected => :single do
+    relevant_mutations = self.relevant_mutations
+
+    samples_with_gene_affected = {}
+    cohort.each do |genotype|
+      genotype.each do |mutation|
+        next unless relevant_mutations.include? mutation
+        genes = []
+        mis = mutation.mutated_isoforms
+        genes.concat mis.select_by(:consequence){|c| c != "SYNONYMOUS"}.transcript.gene unless mis.nil? or mis.empty?
+        genes.concat mutation.transcripts_with_affected_splicing.gene
+        genes.uniq.each{|gene| samples_with_gene_affected[gene] ||= []; samples_with_gene_affected[gene] << genotype.jobname}
+      end
+    end
+    samples_with_gene_affected
+  end
+
+  property :gene_sample_matrix => :single do
+    genotyped_samples = samples.select{|s| s.has_genotype?}.sort.uniq
+
+    tsv = TSV.setup({}, :key_field => "Ensembl Gene ID", :namespace => organism, :type => :list, :fields => genotyped_samples)
+
+    num_samples = genotyped_samples.length
+    genotyped_samples.each_with_index do |sample,i|
+      affected_genes = sample.affected_genes
+      next if affected_genes.empty?
+      affected_genes.clean_annotations.each do |gene|
+        tsv[gene] ||= ["FALSE"] * num_samples
+        tsv[gene][i] = "TRUE"
+      end
+    end
+
+    tsv.fields = genotyped_samples
+
+    tsv
+  end
+
+  property :recurrent_genes => :single do |*args|
+    min = args.first 
+    min = 2 if min.nil?
+
+    Gene.setup(samples_with_gene_affected.select{|gene, samples| samples.length >= min }.collect{|gene,samples| gene}, "Ensembl Gene ID", organism)
+  end
+end

data/lib/rbbt/entity/study/genotypes/knowledge_base.rb

@@ -0,0 +1,39 @@
+module Study
+
+  self.study_registry[:mutation_genes] = Proc.new{|study,database|
+    tsv = TSV.setup({}, :key_field => "Genomic Mutation", :fields => ["Ensembl Gene ID"], :type => :flat, :namespace => study.organism)
+    study.cohort.metagenotype.uniq.each do |mutation|
+      tsv[mutation] = mutation.genes
+    end
+    tsv
+  }
+
+  self.study_registry[:sample_mutations] = Proc.new{|study,database|
+    tsv = TSV.setup({}, :key_field => "Sample", :fields => ["Genomic Mutation"], :type => :flat, :namespace => study.organism)
+    study.samples.select_by(:has_genotype?).each do |sample|
+      tsv[sample] = sample.mutations
+    end
+    tsv
+  }
+
+  self.study_registry[:sample_genes] = Proc.new{|study,database|
+    tsv = TSV.setup({}, :key_field => "Sample", :fields => ["Ensembl Gene ID"], :type => :flat, :namespace => study.organism)
+    study.samples.select_by(:has_genotype?).each do |sample|
+      tsv[sample] = sample.affected_genes
+    end
+    tsv
+  }
+
+  self.study_registry[:sample_genes2] = Proc.new{|study,database|
+    tsv = TSV.setup({}, :key_field => "Sample", :fields => ["Ensembl Gene ID", "Genomic Mutation"], :type => :double, :namespace => study.organism)
+    kb = study.knowledge_base.get_database(:mutation_genes, :source => "Ensembl Gene ID")
+    study.samples.select_by(:has_genotype?).each do |sample|
+      values = sample.affected_genes.collect do |gene|
+        [gene, kb[gene] * ";;"]
+      end
+      tsv[sample] = Misc.zip_fields values
+    end
+    tsv
+  }
+
+end

data/lib/rbbt/entity/study/genotypes/mutations.rb

@@ -0,0 +1,34 @@
+module Study
+  property :all_mutations do
+    cohort.metagenotype.tap{|o| o.jobname = "All mutations in #{ self }" }
+  end
+
+  property :relevant_mutations do
+    all_mutations = self.all_mutations
+
+    all_mutations.select_by(:relevant?).tap{|o| o.jobname = "Relevant mutations in #{ self }" }
+  end
+
+  property :damaging_mutations do |*args|
+    relevant_mutations.select_by(:damaging?, *args).tap{|o| o.jobname = "Damaging mutations in #{ self }" }
+  end
+
+  property :mutations_altering_isoform_sequence do
+    relevant_mutations.select{|m| 
+      mis = m.mutated_isoforms; not mis.nil? and mis.select{|m| m.consequence != "SYNONYMOUS"}.any?
+    }.tap{|o| o.jobname = "Mutations altering isoform sequence in #{ self }"}
+  end
+
+  property :mutations_affecting_splicing_sites do
+    relevant_mutations.select_by(:transcripts_with_affected_splicing){|ts| ts.any? }.
+      tap{|o| o.jobname = "Mutations affecting splicing sites in #{ self }"}
+  end
+
+  property :mutations_over_gene do |gene|
+    all_mutations.select_by(:genes){|genes| genes and genes.include? gene}
+  end
+
+  property :mutations_over_gene_list do |list|
+    all_mutations.select_by(:genes){|genes| genes and (genes & list).any?}
+  end
+end

data/lib/rbbt/entity/study/genotypes/samples.rb

@@ -0,0 +1,28 @@
+module Sample
+  property :has_genotype? => :array2single do
+    study.cohort.values_at(*self).collect{|g| not g.nil?}
+  end
+
+  property :mutations do
+    Study.setup(study)
+    study.cohort[self]
+  end
+
+  property :relevant_mutations do
+    mutations.select_by(:relevant?)
+  end
+
+  property :damaging_mutations do |*args|
+    mutations.select_by(:damaging?, *args)
+  end
+
+  property :affected_genes do
+    mutations.affected_genes.compact.flatten.uniq
+  end
+
+  property :damaged_genes do |*args|
+    mutations.damaged_genes(*args).compact.flatten.uniq
+  end
+end
+
+

data/lib/rbbt/entity/study/genotypes.rb

@@ -0,0 +1,110 @@
+require 'rbbt/entity/genotype'
+
+require 'rbbt/entity/study/genotypes/samples'
+require 'rbbt/entity/study/genotypes/mutations'
+require 'rbbt/entity/study/genotypes/genes'
+require 'rbbt/entity/study/genotypes/enrichment'
+require 'rbbt/entity/study/genotypes/knowledge_base'
+
+Workflow.require_workflow "NKIWorkflow"
+Workflow.require_workflow "TSVWorkflow"
+
+module StudyWorkflow
+  helper :organism do
+    study.metadata[:organism]
+  end
+
+  task :genotype_overview => :tsv do
+    gene_overview = TSV.setup({}, 
+                          :key_field => "Ensembl Gene ID", 
+                          :fields => ["Samples with gene mutated", "Samples with gene affected", "Samples with gene damaged", "Mutation significance"],
+                          :type => :double
+                         )
+    genotyped_samples = study.samples.select_by(:has_genotype?)
+    all_mutations = study.all_mutations
+    if all_mutations.empty?
+      gene_overview 
+    else
+
+      log :affected_genes, "Computing how genes are affected by mutations"
+      mutation_genes = Misc.process_to_hash(all_mutations){|all_mutations| all_mutations.genes}
+      mutation_affected_genes = Misc.process_to_hash(all_mutations){|all_mutations| all_mutations.affected_genes}
+      if all_mutations.length < 5000
+        log :damaged_genes, "Computing genes damaged genes"
+        mutation_damaged_genes = Misc.process_to_hash(all_mutations){|all_mutations| all_mutations.damaged_genes}
+      else
+        mutation_damaged_genes = Misc.process_to_hash(all_mutations){|all_mutations| [nil] * all_mutations.length}
+      end
+      log :significance, "Computing mutation significance"
+      mutation_significance = NKIWorkflow.job(:significantly_mutated, study, :study => study, :threshold => 0.1).run
+      log :significance, "Reordering mutation significance file"
+
+      #TSVWorkflow.job(:change_id, study, :format => "Ensembl Gene ID", :tsv => mutation_significance).run
+      mutation_significance.identifiers = Organism.identifiers(study.organism)
+      mutation_significance = mutation_significance.change_key "Ensembl Gene ID" 
+
+      log :samples, "Gathering affected samples"
+      samples_gene_status = {}
+      genotyped_samples.each do |sample|
+        samples_gene_status[sample] = {}
+
+        mutation_genes.values_at(*sample.mutations).each do |genes|
+          genes.each do |gene|
+            samples_gene_status[sample][gene] ||= [false, false, false]
+            samples_gene_status[sample][gene][0] = true
+          end
+        end
+
+        mutation_affected_genes.values_at(*sample.mutations).each do |genes|
+          genes.each do |gene|
+            samples_gene_status[sample][gene] ||= [false, false, false]
+            samples_gene_status[sample][gene][1] = true
+          end
+        end
+
+        mutation_damaged_genes.values_at(*sample.mutations).each do |genes|
+          next if genes.nil?
+          genes.each do |gene|
+            samples_gene_status[sample][gene] ||= [false, false, false]
+            samples_gene_status[sample][gene][2] = true
+          end 
+        end
+      end
+
+      log :compiling, "Compiling result"
+      mutation_genes.values.compact.flatten.uniq.each do |gene|
+        gene_overview[gene] = []
+        gene_overview[gene] << samples_gene_status.select{|sample, gene_status| gene_status.include? gene and gene_status[gene][0]}.collect{|sample, gene_status| sample}
+        gene_overview[gene] << samples_gene_status.select{|sample, gene_status| gene_status.include? gene and gene_status[gene][1]}.collect{|sample, gene_status| sample} 
+        gene_overview[gene] << samples_gene_status.select{|sample, gene_status| gene_status.include? gene and gene_status[gene][2]}.collect{|sample, gene_status| sample} 
+        gene_overview[gene] << [mutation_significance.include?(gene) ? mutation_significance[gene]["p.value"] : "> 0.1"]
+      end
+
+      gene_overview
+    end
+  end
+end
+
+module Study
+  def has_genotypes?
+    dir.genotypes.exists?
+  end
+
+  attr_accessor :watson
+  def watson
+    @watson  = metadata[:watson] if @watson.nil?
+    @watson
+  end
+
+  def genotype_files
+    dir.genotypes.glob("*")
+  end
+
+  def cohort
+    @cohort ||= genotype_files.collect do |f| 
+      name = File.basename(f)
+      genomic_mutations = Open.read(f).split("\n").sort
+      GenomicMutation.setup(genomic_mutations, name, organism, watson)
+    end.tap{|cohort| cohort.extend Genotype::Cohort}
+  end
+end

data/lib/rbbt/entity/study/knowledge_base.rb

@@ -0,0 +1,36 @@
+module Study
+
+  class << self
+    attr_accessor :knowledge_base, :study_registry
+    def knowledge_base
+      @knowledge_base ||= KnowledgeBase.new Rbbt.var.knowledge_base.Study 
+    end
+
+    def study_registry
+      @study_registry ||= {}
+    end
+  end
+
+  attr_accessor :knowledge_base
+
+  def knowledge_base
+    @knowledge_base ||= begin 
+                          kb = KnowledgeBase.new(self.dir.var.knowledge_base, self.organism)
+                          kb.format["Gene"] = "Ensembl Gene ID"
+                          kb.entity_options["Sample"] = {"Study" => self}
+                          Study.study_registry.each do |database, file|
+                            Log.debug("Inheriting #{ database } from registry: #{Misc.fingerprint file}")
+                            if Proc === file
+                              study = self
+                              block = Proc.new{ file.call(self, database) }
+                              block.define_singleton_method(:filename) do [database, study] * "@" end
+                              kb.register database, nil, {}, &block 
+                            else
+                              kb.register database, file
+                            end
+                          end
+                          kb
+                        end
+  end
+end
+

data/lib/rbbt/entity/study/methylation/samples.rb

@@ -0,0 +1,31 @@
+module Sample
+  property :methylation => :array2single do
+    study.methylation_cohort
+  end
+
+  property :has_methylation? => :array2single do
+    study.methylation_cohort.values_at(*self).collect{|methylation| not methylation.nil?}
+  end
+  
+  property :methylated => :single do
+    return [] if methylation.empty?
+    methylation.select_by(:methylated?)
+  end
+ 
+  property :unmethylated => :single do
+    return [] if methylation.empty?
+    methylation.select_by(:unmethylated?)
+  end
+
+  property :methylated_genes => :single do
+    return [] if methylated.empty?
+    Gene.setup(methylated.genes.flatten.uniq, "Ensembl Gene ID", organism)
+  end
+
+  property :unmethylated_genes => :single do
+    return [] if unmethylated.empty?
+    Gene.setup(unmethylated.genes.flatten.uniq, "Ensembl Gene ID", organism)
+  end
+end
+
+

data/lib/rbbt/entity/study/methylation.rb

@@ -0,0 +1,90 @@
+require 'rbbt/entity/methylation'
+
+require 'rbbt/entity/study/methylation/samples'
+
+module StudyWorkflow
+  helper :organism do
+    study.metadata[:organism]
+  end
+end
+
+module Study
+  def has_methylation?
+    dir.methylation.exists?
+  end
+
+  def methylation_files
+    dir.methylation.find.glob("*")
+  end
+
+  def methylation_cohort
+    if @methylation_cohort.nil?
+      @methylation_cohort = {}
+      methylation_files.each do |f| 
+        sample = File.basename(f)
+        Sample.setup(sample, self)
+        methylations = Open.read(f).split("\n").sort
+        Methylation.setup(methylations, organism)
+        @methylation_cohort[sample] =  methylations
+      end
+    end
+    @methylation_cohort
+  end
+end
+
+module Study
+  property :recurrently_lost_genes => :single do |threshold|
+    counts = {}
+    self.samples.each do |sample|
+      next unless sample.has_methylation?
+      puts sample
+
+      genes = nil
+      genes = sample.lost_genes.clean_annotations
+      genes.each do |gene|
+        counts[gene] ||= 0
+        counts[gene] += 1 
+      end
+    end
+
+    recurrent = counts.select{|k,c| c >= threshold }.collect{|k,v| k }
+    Gene.setup(recurrent, "Ensembl Gene ID", organism)
+  end
+
+  property :recurrently_gained_genes => :single do |threshold|
+    counts = {}
+    self.samples.each do |sample|
+      next unless sample.has_methylation?
+      puts sample
+
+      genes = nil
+      genes = sample.gained_genes.clean_annotations
+      genes.each do |gene|
+        counts[gene] ||= 0
+        counts[gene] += 1 
+      end
+    end
+
+    recurrent = counts.select{|k,c| c >= threshold }.collect{|k,v| k }
+    Gene.setup(recurrent, "Ensembl Gene ID", organism)
+  end
+
+  property :gene_sample_methylation_matrix => :single do
+    tsv = TSV.setup({}, :key_field => "Ensembl Gene ID", :namespace => organism, :type => :list)
+    samples = []
+    i = 0
+    num_samples = cohort.length
+    methylation_cohort.each do |sample,methylation|
+      methylation.genes.compact.flatten.uniq.each do |gene|
+        tsv[gene] ||= ["FALSE"] * num_samples
+        tsv[gene][i] = "TRUE"
+      end
+      samples << sample
+      i += 1
+    end
+
+    tsv.fields = samples
+
+    tsv
+  end
+end