Module: IDConverter::Light

Defined in:
lib/idclight.rb

Constant Summary collapse

@@HUMAN =

Organism type constants

'Hs'
@@MOUSE = 
'Mm'
@@RAT = 
'Rn'
@@UNIGENE =

ID type constants.

'ug'
@@ENTREZ = 
'entrez'
@@ENSEMBL = 
'ensembl'
@@HUGO = 
'hugo'
@@GENBANK = 
'acc'
@@CLONE = 
'clone'
@@AFFYMETRIX = 
'affy'
@@REFSEQ_RNA = 
'rsdna'
@@REFSEQ_PEPTIDE = 
'rspep'
@@SWISSPROT = 
'swissp'

Instance Method Summary collapse

Instance Method Details

#ensembl_id_to_entrez_id(id, organism = @@HUMAN) ⇒ Object 



79
80
81
82
 
# File 'lib/idclight.rb', line 79

def ensembl_id_to_entrez_id(id, organism = @@HUMAN)
 	data = search(@@ENSEMBL, id, organism)
 		extract_entrez_id(data)
end

#ensembl_id_to_hugo_id(id, organism = @@HUMAN) ⇒ Object 



84
85
86
87
 
# File 'lib/idclight.rb', line 84

def ensembl_id_to_hugo_id(id, organism = @@HUMAN)
 	data = search(@@ENSEMBL, id, organism)
 		extract_hugo_id(data)
end

#extract_ensembl_id(data) ⇒ Object 



107
108
109
110
 
# File 'lib/idclight.rb', line 107

def extract_ensembl_id(data)
	ent = data['Ensembl']
	extract_last_link_text(ent)
end

#extract_entrez_id(data) ⇒ Object 



89
90
91
92
93
94
95
96
97
98
99
100
 
# File 'lib/idclight.rb', line 89

def extract_entrez_id(data)
	ent = data['Entrez Gene']
	id = nil
	if ent
		doc = Hpricot(ent)
        doc.search('//a').each do |e|
          tmp = e.inner_html.to_i
          id = tmp if tmp != 0 
        end
	end	
	id
end

#extract_hugo_id(data) ⇒ Object 



102
103
104
105
 
# File 'lib/idclight.rb', line 102

def extract_hugo_id(data)
	ent = data['HUGO']
	extract_last_link_text(ent)
end

#extract_last_link_text(ent) ⇒ Object 



112
113
114
115
116
117
118
119
120
121
122
 
# File 'lib/idclight.rb', line 112

def extract_last_link_text(ent)
	id = nil
	if ent
		doc = Hpricot(ent)
        doc.search('//a').each do |e|
          tmp = e.inner_html
          id = tmp if !tmp.nil? && tmp != ''
        end
	end	
	id
end

#hugo_id_to_ensembl_id(id, organism = @@HUMAN) ⇒ Object 



74
75
76
77
 
# File 'lib/idclight.rb', line 74

def hugo_id_to_ensembl_id(id, organism = @@HUMAN)
  data = search(@@HUGO, id, organism)
  extract_ensembl_id(data)
end

#hugo_id_to_entrez_id(id, organism = @@HUMAN) ⇒ Object 



69
70
71
72
 
# File 'lib/idclight.rb', line 69

def hugo_id_to_entrez_id(id, organism = @@HUMAN)
  data = search(@@HUGO, id, organism)
  extract_entrez_id(data)
end

#search(id_type, id, organism = @@HUMAN) ⇒ Object

Originally taken from idclight.bioinfo.cnio.es/ idtype refers to the type of ID from which you want to obtain further information. Current options are: ug (UniGene cluster) entrez (EntrezGene ID) ensembl (Ensembl Gene) hugo (HUGO Gene Name) acc (GenBank accession) clone (Clone ID) affy (Affymetrix ID) rsdna (RefSeq_RNA) rspep (RefSeq_peptide) swissp (SwissProt name) id is the ID of the gene or clone for which more information is required. org is the organism you are working with. Three different organisms are available Hs (Human - Homo sapiens), Mm (Mouse - Mus musculus), and Rn (Rat - Rattus norvegicus).



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
 
# File 'lib/idclight.rb', line 42

def search(id_type, id, organism = @@HUMAN)
  return nil if id.nil? || id_type.nil? || organism.nil?
  url = "http://idclight.bioinfo.cnio.es/idclight.prog?id=#{id}&idtype=#{id_type}&org=#{organism}"
  data = nil
  
  begin
    doc = Hpricot(URI.parse(url).read)
    elements = doc.search("//table[@class='sample']//tr")
    data = {}
    # last = nil
    elements.each do |e|
      tds = e/'td'
      tds.size == 2
      t1 = tds[0].inner_html
      t2 = tds[1].inner_html
      if t1.nil? || t1 == ''
        # TODO
      else
        data[t1] = t2
      end
    end
  rescue SocketError => e
    data =nil
  end
  data
end