Sha256: 74fe332db8574851c57b092ceb34dc9da774cd0ab7865c8093639d2e0b409747
Contents?: true
Size: 1.88 KB
Versions: 4
Compression:
Stored size: 1.88 KB
Contents
require 'rbbt/util/open'
require 'rbbt/util/misc'
def read_chunk(jochem)
chunk = ""
while (not jochem.eof? and not (line = jochem.gets).match(/^--/))
chunk << line
end
return nil if chunk.empty?
chunk
end
def first(list)
return nil if list.nil? or list.empty?
list.first
end
def process_jochem
jochem = Open.open("http://www.biosemantics.org/uploads/file/Jochem/JochemV1_2.zip")
identifiers = File.open('identifiers', 'w')
identifiers.puts("#: :namespace=JoChem")
identifiers.puts("#ID\tCompound Name\tPubChem:ID\tDrugBank:ID")
lexicon = File.open('lexicon', 'w')
lexicon.puts("#: :namespace=JoChem")
lexicon.puts("#ID\tSynonyms")
inchi = File.open('inchi', 'w')
inchi.puts("#: :namespace=JoChem")
inchi.puts("#ID\tInChi")
definitions = File.open('definitions', 'w')
definitions.puts("#: :namespace=JoChem#:type=:list")
definitions.puts("#ID\tDefinition")
while chunk = read_chunk(jochem) do
next if chunk.empty? or chunk =~ /^#/ or chunk =~ /^NS /
info = {}
chunk.split(/\n/).each do |line|
line.sub!(/\t@match.*/,'')
code, value = line.match(/([A-Z]*) (.*)/).values_at 1, 2
info[code] ||= []
info[code] << value
end
id = first(info["ID"])
na = first(info["NA"])
df = first(info["DF"])
tm = info["TM"] || []
db = info["DB"] || []
pubc = db.collect{|code| code.match(/PUBC_(.*)/) ? $1 : nil}.compact
drug = db.collect{|code| code.match(/DRUG_(.*)/) ? $1 : nil}.compact
inch = db.collect{|code| code.match(/INCH_InChI=(.*)/) ? $1 : nil}.compact
lexicon.puts [id, tm.unshift(na) * "|"] * "\t"
identifiers.puts [id, na, pubc * "|", drug * "|"] * "\t"
inchi.puts [id, inch * "|"] * "\t" if inch.any?
definitions.puts [id, df] * "\t" unless df.nil?
end
end
rule /identifiers|lexicon|inchi|definitions/ do |t|
Misc.in_dir(File.dirname(t.name)) do
process_jochem
end
end
Version data entries
4 entries across 4 versions & 1 rubygems