Module: Study

Extended by:
Entity, Resource
Includes:
LocalPersist
Defined in:
lib/rbbt/entity/study.rb,
lib/rbbt/entity/study/cnv.rb,
lib/rbbt/entity/study/cnv.rb,
lib/rbbt/entity/study/snp.rb,
lib/rbbt/entity/study/snp.rb,
lib/rbbt/entity/study/samples.rb,
lib/rbbt/entity/study/samples.rb,
lib/rbbt/entity/study/cnv/genes.rb,
lib/rbbt/entity/study/genotypes.rb,
lib/rbbt/entity/study/expression.rb,
lib/rbbt/entity/study/methylation.rb,
lib/rbbt/entity/study/methylation.rb,
lib/rbbt/entity/study/knowledge_base.rb,
lib/rbbt/entity/study/genotypes/genes.rb,
lib/rbbt/entity/study/cnv/knowledge_base.rb,
lib/rbbt/entity/study/genotypes/mutations.rb,
lib/rbbt/entity/study/genotypes/knowledge_base.rb

Class Attribute Summary collapse

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Class Attribute Details

.knowledge_base ⇒ Object

Returns the value of attribute knowledge_base.



4
5
6
 
# File 'lib/rbbt/entity/study/knowledge_base.rb', line 4

def knowledge_base
  @knowledge_base
end

.study_dir ⇒ Object

Returns the value of attribute study_dir.



50
51
52
 
# File 'lib/rbbt/entity/study.rb', line 50

def study_dir
  @study_dir
end

.study_registry ⇒ Object

Returns the value of attribute study_registry.



4
5
6
 
# File 'lib/rbbt/entity/study/knowledge_base.rb', line 4

def study_registry
  @study_registry
end

Instance Attribute Details

#dir ⇒ Object

Returns the value of attribute dir.



65
66
67
 
# File 'lib/rbbt/entity/study.rb', line 65

def dir
  @dir
end

#knowledge_base ⇒ Object

Returns the value of attribute knowledge_base.



14
15
16
 
# File 'lib/rbbt/entity/study/knowledge_base.rb', line 14

def knowledge_base
  @knowledge_base
end

#organism ⇒ Object

{{{ Attributes



144
145
146
 
# File 'lib/rbbt/entity/study.rb', line 144

def organism
  @organism
end

#watson ⇒ Object

Returns the value of attribute watson.



119
120
121
 
# File 'lib/rbbt/entity/study/genotypes.rb', line 119

def watson
  @watson
end

#workflow(&block) ⇒ Object

Returns the value of attribute workflow.



65
66
67
 
# File 'lib/rbbt/entity/study.rb', line 65

def workflow
  @workflow
end

Class Method Details

.annotation_repo ⇒ Object 



87
88
89
 
# File 'lib/rbbt/entity/study.rb', line 87

def self.annotation_repo
  @annotation_repo ||= Rbbt.var.cache.annotation_repo.find
end

.extended(base) ⇒ Object 



91
92
93
94
95
96
97
98
99
100
101
102
 
# File 'lib/rbbt/entity/study.rb', line 91

def self.extended(base)
  base.workflow = StudyWorkflow.clone
  base.workflow.study = base

  if File.exists?(setup_file = File.join(base.dir, 'setup.rb'))
    base.instance_eval Open.read(setup_file), setup_file
  end

  base.local_persist_dir = Rbbt.var.cache.studies[base].persistence.find

  base
end

.identify_study(samples) ⇒ Object 



71
72
73
74
75
76
77
 
# File 'lib/rbbt/entity/study/samples.rb', line 71

def self.identify_study(samples)
  samples = Array === samples ? samples.flatten : [samples]

  studies = Study.studies.select{|study| Study.setup(study); (study.samples & samples).any? }

  studies.first
end

.studies ⇒ Object 



104
105
106
107
 
# File 'lib/rbbt/entity/study.rb', line 104

def self.studies
  Dir.glob(File.join(Path === study_dir ? study_dir.find : study_dir, '*')).
    select{|f| File.directory? f}.sort.collect{|s| Study.setup(File.basename(s))}
end

Instance Method Details

#cnv_cohort ⇒ Object 



61
62
63
64
65
66
67
68
69
70
71
72
73
 
# File 'lib/rbbt/entity/study/cnv.rb', line 61

def cnv_cohort
  if @cnv_cohort.nil?
    @cnv_cohort = {}
    cnv_files.each do |f| 
      sample = File.basename(f)
      Sample.setup(sample, self)
      cnvs = Open.read(f).split("\n").sort
      CNV.setup(cnvs, organism)
      @cnv_cohort[sample] =  cnvs
    end
  end
  @cnv_cohort
end

#cnv_files ⇒ Object 



57
58
59
 
# File 'lib/rbbt/entity/study/cnv.rb', line 57

def cnv_files
  dir.cnv.find.glob("*")
end

#cohort ⇒ Object 



129
130
131
132
133
134
135
 
# File 'lib/rbbt/entity/study/genotypes.rb', line 129

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

#genotype_files ⇒ Object 



125
126
127
 
# File 'lib/rbbt/entity/study/genotypes.rb', line 125

def genotype_files
  dir.genotypes.glob("*")
end

#has_cnv? ⇒ Boolean 



53
54
55
 
# File 'lib/rbbt/entity/study/cnv.rb', line 53

def has_cnv?
  dir.cnv.exists?
end

#has_expression? ⇒ Boolean 



21
22
23
 
# File 'lib/rbbt/entity/study/expression.rb', line 21

def has_expression?
  dir.matrices["gene_expression"].exists?
end

#has_genotypes? ⇒ Boolean 



115
116
117
 
# File 'lib/rbbt/entity/study/genotypes.rb', line 115

def has_genotypes?
  dir.genotypes.exists?
end

#has_methylation? ⇒ Boolean 



12
13
14
 
# File 'lib/rbbt/entity/study/methylation.rb', line 12

def has_methylation?
  dir.methylation.exists?
end

#has_snp? ⇒ Boolean 



9
10
11
 
# File 'lib/rbbt/entity/study/snp.rb', line 9

def has_snp?
  dir.snp.exists?
end

#job(task, *args) ⇒ Object 



67
68
69
70
71
72
73
74
75
76
77
 
# File 'lib/rbbt/entity/study.rb', line 67

def job(task, *args)
  name, inputs = args
  if inputs.nil? and Hash === name
    inputs = name
    name = nil
  end
  name = self if name.nil? or name == :self or name == "self"
  step = workflow.job(task, name, {:organism => metadata[:organism], :watson => metadata[:watson]}.merge(inputs || {}))
  step.instance_variable_set(:@study, self)
  step
end

#match_samples(list) ⇒ Object 



62
63
64
65
66
67
68
69
 
# File 'lib/rbbt/entity/study/samples.rb', line 62

def match_samples(list)
  if donor_id_field = (sample_info = self.sample_info).fields.select{|f| f =~ /donor\s+id/i}.first
    list_donors = sample_info.select(list).slice(donor_id_field).values.compact.flatten
    list_donor_samples = sample_info.select(list_donors).keys
    list = list_donor_samples.annotate((list + list_donor_samples).uniq)
  end
  list
end

#matrices ⇒ Object 



153
154
155
 
# File 'lib/rbbt/entity/study.rb', line 153

def matrices
  dir.matrices.glob('*').collect{|f| f.basename}
end

#matrix(type, format = "Ensembl Gene ID", organism = nil) ⇒ Object 



157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
 
# File 'lib/rbbt/entity/study.rb', line 157

def matrix(type, format = "Ensembl Gene ID", organism = nil)
  organism = self.metadata[:organism] if organism.nil?
  raise "No matrices defined for study #{ self }" unless defined? matrices.empty?
  raise "No type specified" if type.nil?
  type = type.to_s
  raise "No matrix #{ type } defined for study #{ self }" unless matrices.include? type
  data = dir.matrices[type].data.find if dir.matrices[type].data.exists?
  if dir.matrices[type].identifiers.exists?
    identifiers = dir.matrices[type].identifiers.find 
  else
    identifiers = Organism.identifiers(organism).find
  end
  samples = dir.matrices[type].samples.find if dir.matrices[type].samples.exists?
  samples = dir.samples.find if samples.nil? and dir.samples.exist?
  Matrix.new(data, identifiers, samples, format, organism)
end

#matrix_file(name) ⇒ Object 



149
150
151
 
# File 'lib/rbbt/entity/study.rb', line 149

def matrix_file(name)
  dir.matrices[name.to_s].produce.find
end

#metadata ⇒ Object 



135
136
137
 
# File 'lib/rbbt/entity/study.rb', line 135

def metadata
  @metadata ||= (dir["metadata.yaml"].yaml.extend IndiferentHash)
end

#methylation_cohort ⇒ Object 



20
21
22
23
24
25
26
27
28
29
30
31
32
 
# File 'lib/rbbt/entity/study/methylation.rb', line 20

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

#methylation_files ⇒ Object 



16
17
18
 
# File 'lib/rbbt/entity/study/methylation.rb', line 16

def methylation_files
  dir.methylation.find.glob("*")
end

#sample_info ⇒ Object 



42
43
44
45
 
# File 'lib/rbbt/entity/study/samples.rb', line 42

def sample_info
  return nil unless dir.samples.exists?
  @sample_info ||= dir.samples.tsv.tap{|tsv| tsv.entity_options = {:study => self }}
end

#samples ⇒ Object 



47
48
49
50
51
52
53
54
55
56
57
58
59
60
 
# File 'lib/rbbt/entity/study/samples.rb', line 47

def samples
  @samples ||= begin
                 samples = local_persist("Sample", :array) do
                   if sample_info.nil?
                     self.cohort.collect{|g| g.jobname }
                   else
                     sample_info.keys
                   end
                 end
                 Sample.setup(samples, :study => self)
                 samples.study = self
                 samples
               end
end

#snp_cohort ⇒ Object 



17
18
19
20
21
22
23
24
25
26
27
28
29
 
# File 'lib/rbbt/entity/study/snp.rb', line 17

def snp_cohort
  if @snp_cohort.nil?
    @snp_cohort = {}
    snp_files.each do |f| 
      sample = File.basename(f)
      Sample.setup(sample, self)
      snps = Open.read(f).split("\n").sort
      SNP.setup(snps)
      @snp_cohort[sample] =  snps
    end
  end
  @snp_cohort
end

#snp_files ⇒ Object 



13
14
15
 
# File 'lib/rbbt/entity/study/snp.rb', line 13

def snp_files
  @snp_files ||= dir.snp.find.glob("*")
end

#snp_index ⇒ Object 



34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
 
# File 'lib/rbbt/entity/study/snp.rb', line 34

def snp_index
  local_persist_tsv("SNP2Samples", "SNP2Samples", {}, :persist => true, :serializer => :clean) do |data|

    require 'progress-monitor'
    Progress.monitor "SNP files", :stack_depth => 0
    snp_files.each do |file|
      file = file.to_s
      sample = File.basename file
      File.open(file.to_s) do |f|
        while line = f.gets
          snp = line.strip
          snp, allele = snp.split ":"
          snp_str = data[snp]

          if snp_str.nil?
            snp_str = ""
          else
            snp_str += "\t"
          end

          if allele
            snp_str << sample << ":" << allele
          else
            snp_str << sample
          end
          data[snp] = snp_str
        end
      end
    end

    TSV.setup data
    data.key_field = "RS ID"
    data.fields = ["Sample"]
    data.type = :flat
    data.serializer = :list
    data
  end
end

#users ⇒ Object 



139
140
141
 
# File 'lib/rbbt/entity/study.rb', line 139

def users
  @users ||= metadata[:users] || []
end