rbbt-study 0.2.30 → 0.2.31

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

@@ -1,34 +0,0 @@
-module Study
-  property :all_mutations do
-    cohort.metagenotype.sort.tap{|o| o.jobname = "All mutations in #{ self }"; o.organism ||= organism; o.watson ||= watson }
-  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

@@ -1,28 +0,0 @@
-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/knowledge_base.rb

@@ -1,35 +0,0 @@
-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(Rbbt.var.knowledge_base.studies[self], 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) }
-                              kb.register database, nil, {}, &block 
-                            else
-                              kb.register database, file
-                            end
-                          end
-                          kb
-                        end
-  end
-end
-

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

@@ -1,90 +0,0 @@
-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

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

@@ -1,31 +0,0 @@
-module Sample
-  property :methylation => :array2single do
-    study.has_methylation? ? study.methylation_cohort : {}
-  end
-
-  property :has_methylation? => :array2single do
-    study.has_methylation? ? study.methylation_cohort.values_at(*self).collect{|methylation| not methylation.nil?} : [false] * self.length
-  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/mutations.rb

@@ -1,259 +0,0 @@
-task :mutations_by_change => :tsv do
-  changes = {}
-
-  study.cohort.each do |genotype|
-    genotype.watson ||= watson
-    genotype.each do |mutation|
-      reference = watson ? mutation.reference : mutation.gene_strand_reference
-      base = mutation.base
-      base = ((Misc::IUPAC2BASE[base] || []) - [reference]) * ","
-      change = [reference, base]
-      changes[change * ">"] ||= []
-      changes[change * ">"] << mutation.clean_annotations
-    end
-  end
-
-  TSV.setup(changes, :key_field => "Genomic Change", :fields => ["Genomic Mutation"], :namespace => organism, :type => :flat)
-
-  changes.entity_options = {:watson => watson}
-
-  changes
-end
-
-
-dep :mutations_by_change
-task :mutation_change_counts => :yaml do
-  change_counts    = {}
-
-  step(:mutations_by_change).load.each do |change, mutations|
-    change_counts[change] = mutations.length
-  end
-
-  change_counts
-end
-
-returns "Genomic Mutation"
-task :transversions => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation|
-
-      mutation.type == "transversion"
-
-    } 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: transversions", organism, watson)
-
-end
-
-returns "Genomic Mutation"
-task :transitions => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation|
-
-      mutation.type == "transition"
-
-    } 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: transitions", organism, watson)
-
-end
-
-returns "Genomic Mutation"
-task :indels => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation|
-
-      mutation.type == "indel"
-
-    } 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: indels", organism, watson)
-end
-
-returns "Genomic Mutation"
-task :unknown_mutations => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation|
-
-      mutation.type == "unknown"
-
-    } 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: unknown_mutations", organism, watson)
-end
-
-
-returns "Genomic Mutation"
-task :not_mutations => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation|
-
-      mutation.type == "none"
-
-    } 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: not mutations", organism, watson)
-end
-
-
-
-returns "Genomic Mutation"
-task :non_synonymous_mutations => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation|
-
-      (mutation.mutated_isoforms || [] ).select{|mi| mi.non_synonymous }.any? 
-
-    } 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: non_synonymous mutations", organism, watson)
-end
-
-dep :non_synonymous_mutations
-returns "Genomic Mutation"
-task :synonymous_mutations => :annotations do
-  non_synonymous_mutations = step(:non_synonymous_mutations).load
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.remove( non_synonymous_mutations ) 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: synonymous mutations", organism, watson)
-end
-
-#dep :synonymous_mutations
-#dep :exon_junction_mutations
-#input :methods, :array, "Damage prediction methods", [:sift, :mutation_assessor]
-#returns "Genomic Mutation"
-#task :damaging_mutations => :annotations do |methods|
-#  synonymous_mutations = step(:synonymous_mutations).load
-#  exon_junction_mutations = step(:exon_junction_mutations).load
-#
-#  mutations_to_remove = synonymous_mutations - exon_junction_mutations
-#
-#  mutations = study.cohort.collect{|genotype| 
-#
-#    genotype.remove( mutations_to_remove ).select{|mutation| mutation.damaging?(methods) } 
-#
-#  }.flatten
-#
-#  GenomicMutation.setup(mutations, "#{ study }: damaging mutations", organism, watson)
-#end
-
-dep :relevant_mutations
-input :methods, :array, "Damage prediction methods", [:sift, :mutation_assessor]
-returns "Genomic Mutation"
-task :damaging_mutations => :annotations do |methods|
-  relevant_mutations = step(:relevant_mutations ).load
-
-  mutations = relevant_mutations.select{|mutation| mutation.damaging?(methods) }
-
-  GenomicMutation.setup(mutations, "#{ study }: damaging mutations", organism, watson)
-end
-
-
-dep :damaging_mutations
-dep :relevant_mutations
-input :methods, :array, "Damage prediction methods", [:sift]
-returns "Genomic Mutation"
-task :mutations_missing_predictions => :annotations do |methods|
-  damaging_mutations = step(:damaging_mutations).load
-  relevant_mutations = step(:relevant_mutations).load
-
-  missing_mutations = relevant_mutations.remove(damaging_mutations)
-  missing_mutations_mutated_isoforms = missing_mutations.mutated_isoforms.compact.flatten
-  mutated_isoforms_missing_damage_scores = missing_mutations_mutated_isoforms.select{|mis| mis.damage_scores.nil?}
-  mutations_missing_predictions = missing_mutations.select{|mutation| mutation.mutated_isoforms and mutation.mutated_isoforms.any?}.select{|mutation| mutation.mutated_isoforms.remove(mutated_isoforms_missing_damage_scores).empty?}
-  GenomicMutation.setup(mutations_missing_predictions, "#{ study }: mutations missing predictions", organism, watson)
-end
-
-returns "Genomic Mutation"
-task :exon_junction_mutations => :annotations do
-
-  mutations = study.cohort.collect{|genotype| 
-
-    genotype.select{|mutation| mutation.transcripts_with_affected_splicing.any? and not mutation.type == "none"} 
-
-  }.flatten
-
-  GenomicMutation.setup(mutations, "#{ study }: exon junction mutations", organism, watson)
-end
-
-dep :non_synonymous_mutations
-dep :exon_junction_mutations
-returns "Genomic Mutation"
-task :relevant_mutations => :annotations do
-  non_synonymous_mutations = step(:non_synonymous_mutations).load
-  exon_junction_mutations = step(:exon_junction_mutations).load
-
-  all_relevant_mutations = ( exon_junction_mutations + non_synonymous_mutations.remove(exon_junction_mutations) ).flatten
-
-  GenomicMutation.setup(all_relevant_mutations, "#{ study }: relevant mutations", organism, watson)
-end
-
-dep :relevant_mutations
-returns "Genomic Mutation"
-task :recurrent_mutations => :annotations do
-  relevant_mutations = step(:relevant_mutations).load
-
-  mutations = Misc.counts(relevant_mutations.remove_score).select{|mutation, count| 
-
-    count > 1 
-
-  }.collect{|mutation, count| mutation}
-
-  GenomicMutation.setup(mutations, "#{study}: recurrent mutations", organism, watson)
-end
-
-dep :non_synonymous_mutations
-task :mutations_by_consequence => :yaml do
-  non_synonymous_mutations = step(:non_synonymous_mutations).load
-
-  mutations_by_consequence = {}
-  study.cohort.each do |genotype|
-    genotype.subset(non_synonymous_mutations).each do |mutation|
-      mis = mutation.mutated_isoforms
-      next if mis.nil?
-      consequences = mis.consequence.compact.uniq
-      consequences.each{|consequence| mutations_by_consequence[consequence] ||= []; mutations_by_consequence[consequence] << mutation }
-    end
-  end
-
-  mutations_by_consequence
-end
-%w(missense_mutations nonsense_mutations frameshift_mutations nostop_mutations indel_mutations utr_mutations ).zip(
-  %w(MISS-SENSE NONSENSE FRAMESHIFT NOSTOP INDEL UTR)).each do |task_name, consequence|
-  dep :mutations_by_consequence
-  returns "Genomic Mutation"
-  task task_name => :annotations do
-    mutations_by_consequence = step(:mutations_by_consequence).load
-    GenomicMutation.setup(mutations_by_consequence[consequence] || [], "#{study}: mutations with #{consequence.downcase} isoform mutations", organism, watson)
-  end
-end
-