Module: Organism

Extended by:
Resource
Defined in:
lib/rbbt/sources/organism.rb

Defined Under Namespace

Classes: OrganismNotProcessedError

Class Method Summary collapse

Class Method Details

.attach_translations(org, tsv, target = nil, fields = nil, options = {}) ⇒ Object 



84
85
86
87
88
89
90
91
92
93
94
 
# File 'lib/rbbt/sources/organism.rb', line 84

def self.attach_translations(org, tsv, target = nil, fields = nil, options = {})
  Log.high "Attaching Translations for #{ org.inspect }, target #{target.inspect}, fields #{fields.inspect}"
  options = Misc.add_defaults options, :persist => true, :case_insensitive => false

  options.merge! :key_field    => target unless target.nil?
  options.merge! :fields => fields unless fields.nil?

  index = identifiers(org).tsv options

  tsv.attach index, :fields => [:key], :persist_input => true
end

.entrez_taxid_organism(taxid) ⇒ Object 



145
146
147
148
149
150
151
152
153
 
# File 'lib/rbbt/sources/organism.rb', line 145

def self.entrez_taxid_organism(taxid)
  all_organisms = Organism.installable_organisms

  all_organisms.each do |organism|
    return organism if Organism.entrez_taxids(organism).read.split("\n").include? taxid.to_s
  end

  raise "No organism identified for taxid #{taxid}. Supported organism are: #{all_organisms * ", "}"
end

.guess_id(org, values) ⇒ Object 



122
123
124
125
 
# File 'lib/rbbt/sources/organism.rb', line 122

def self.guess_id(org, values)
  field_matches = TSV.field_match_counts(Organism.identifiers(org).find, values)
  field_matches.sort_by{|field, count| count.to_i}.last
end

.hg_build(organism) ⇒ Object 



22
23
24
25
26
27
28
29
30
31
32
33
 
# File 'lib/rbbt/sources/organism.rb', line 22

def self.hg_build(organism)
  require 'rbbt/sources/ensembl_ftp'

  raise "Only organism 'Hsa' (Homo sapiens) supported" unless organism =~ /^Hsa/

  return 'hg19' unless organism =~ /\//
  date = organism.split("/")[1] 

  release = Ensembl.releases[date]

  release.sub(/.*-/,'').to_i > 54 ? 'hg19' : 'hg18'
end

.installable_organisms ⇒ Object 



9
10
11
 
# File 'lib/rbbt/sources/organism.rb', line 9

def self.installable_organisms
  Rbbt.share.install.Organism.find.glob('???').collect{|f| File.basename(f)}
end

.known_ids(name) ⇒ Object 



141
142
143
 
# File 'lib/rbbt/sources/organism.rb', line 141

def self.known_ids(name)
  TSV::Parser.new(Organism.identifiers(name).open).all_fields
end

.liftOver(positions, source, target) ⇒ Object 



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
72
73
74
75
76
77
78
79
80
 
# File 'lib/rbbt/sources/organism.rb', line 35

def self.liftOver(positions, source, target)

  source_hg = hg_build(source)
  target_hg = hg_build(target)

  case
  when (source_hg == 'hg19' and target_hg == 'hg18')
    map_url = "http://hgdownload.cse.ucsc.edu/goldenPath/hg19/liftOver/hg19ToHg18.over.chain.gz" 
  when (source_hg == 'hg18' and target_hg == 'hg19')
    map_url = "http://hgdownload.cse.ucsc.edu/goldenPath/hg18/liftOver/hg18ToHg19.over.chain.gz" 
  else
    return positions
  end

  positions_bed = positions.collect{|position| 
    chr, pos = position.split(":").values_at(0,1)
    ["chr" << chr, pos.to_i-1, pos, position] * "\t"
  } * "\n" + "\n"

  new_positions = {}

  TmpFile.with_file(positions_bed) do |source_bed|
    TmpFile.with_file() do |unmapped_file|
      TmpFile.with_file() do |map_file|


        Open.write(map_file, Open.read(map_url))
        new_mutations = TmpFile.with_file() do |target_bed|
          FileUtils.chmod(755, Rbbt.software.opt.bin.liftOver.produce.find)
          CMD.cmd("#{Rbbt.software.opt.bin.liftOver.find} '#{source_bed}' '#{map_file}' '#{target_bed}' '#{unmapped_file}'").read
          Open.read(target_bed) do |line|
            chr, position_alt, position, name = line.chomp.split("\t")
            chr.sub! /chr/, ''

            old_chr, old_position, *rest = name.split(":")
            new_positions[name] = ([chr, position].concat rest) * ":"
          end
        end
      end
    end
  end

  positions.collect do |position|
    new_positions[position]
  end
end

.normalize(org, list, target = nil, fields = nil, options = {}) ⇒ Object 



96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
 
# File 'lib/rbbt/sources/organism.rb', line 96

def self.normalize(org, list, target = nil, fields = nil, options = {})
  return [] if list.nil? or list.empty?
  options = Misc.add_defaults options, :persist => true, :case_insensitive => true, :double => false
  double = Misc.process_options options, :double


  options.merge! :target => target unless target.nil?
  options.merge! :fields => fields unless fields.nil?

  index = identifiers(org).index options

  if Array === list
    if double
      index.values_at *list
    else
      index.values_at(*list).collect{|e| Misc.first e}
    end
  else
    if double
      index[list]
    else
      index[list].first
    end
  end
end

.organism(name) ⇒ Object 



135
136
137
138
139
 
# File 'lib/rbbt/sources/organism.rb', line 135

def self.organism(name)
  organisms.select{|organism|
    organism == name or Organism.scientific_name(organism) =~ /#{ name }/i
  }.first
end

.organisms ⇒ Object 



127
128
129
 
# File 'lib/rbbt/sources/organism.rb', line 127

def self.organisms
  Dir.glob(File.join(Organism.root.find, '*')).collect{|f| File.basename(f)}
end

.scientific_name(organism) ⇒ Object 



131
132
133
 
# File 'lib/rbbt/sources/organism.rb', line 131

def self.scientific_name(organism)
  Organism[organism]["scientific_name"].produce.read.strip
end