1 files changed, 29 insertions, 0 deletions
diff --git a/scripts/efsa2csv.rb b/scripts/efsa2csv.rb
new file mode 100755
index 0000000..5251cae
--- /dev/null
+++ b/scripts/efsa2csv.rb
@@ -0,0 +1,29 @@
+#!/usr/bin/env ruby
+require 'csv'
+require_relative '../../lazar/lib/lazar.rb'
+
+i = 0
+db = {}
+CSV.foreach(ARGV[0], :encoding => "UTF-8", :col_sep => "\t", :liberal_parsing => true) do |row|
+  #STDERR.puts i if i%100 == 0
+  if i > 0 and row[11] and !row[11].empty? and row[24].match(/Salmonella/i) and ( row[25].match("TA 98") or row[25].match("TA 100") ) and row[33]
+    begin
+      c = OpenTox::Compound.from_smiles(row[11].gsub('"','')).smiles
+    rescue
+      c = OpenTox::Compound.from_inchi(row[12]).smiles # some smiles (row[11]) contain non-parseable characters
+    end
+    db[c] ||= {}
+    db[c][:id] ||= row[2]
+    if row[33].match(/Positiv/i)
+      db[c][:value] = 1 # at least one positive result in TA 98 or TA 100
+    elsif row[33].match(/Negativ/i)
+      db[c][:value] ||= 0
+    end
+  end
+  i += 1
+end
+
+puts "ID,SMILES,Mutagenicity"
+db.each do |s,v|
+  puts [v[:id],s,v[:value]].join ","
+end