From 36ea2f865014a761be9fe74719f4c88dbaffeb81 Mon Sep 17 00:00:00 2001
From: mguetlein <martin.guetlein@gmail.com>
Date: Mon, 27 Oct 2014 15:48:01 +0100
Subject: inital commit more documentation will be added

---
 nch/03_validate_compounds.rb | 49 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
 create mode 100755 nch/03_validate_compounds.rb

(limited to 'nch/03_validate_compounds.rb')

diff --git a/nch/03_validate_compounds.rb b/nch/03_validate_compounds.rb
new file mode 100755
index 0000000..1ddb719
--- /dev/null
+++ b/nch/03_validate_compounds.rb
@@ -0,0 +1,49 @@
+#!/usr/bin/env ruby
+
+require "csv"
+require "uri"
+
+require "./config.rb"
+
+DATA.each do |d|
+  puts "\ndataset name #{d}"
+
+  next if d=="LOAEL-mg"
+
+  @inchis = []
+
+  all_compounds = (d=="LOAEL-mol" ? "endpoint" : "complete")    
+  [all_compounds,"orig-features"].each do |mode|
+
+    csv_string = CSV.generate({:force_quotes=>true}) do |csv|
+      first = true
+      CSV.foreach("data/02/#{d}_#{mode}.csv") do |row|
+        if (first)
+          first = false
+        else
+          inchi = URI.unescape(row[0])
+          if mode==all_compounds
+            @inchis << inchi
+          else
+            raise "features-set inchi not found in complete-set #{inchi}" unless @inchis.first==inchi
+            @inchis.delete(inchi)
+          end
+        end
+      end
+    end
+    if mode==all_compounds
+      puts @inchis.size.to_s+" compounds in complete-set"
+      @inchis.uniq!
+      File.open("data/03/#{d}_uniq.csv","w").puts "\""+(["InChI"]+@inchis).join("\"\n\"")+"\""
+      puts @inchis.size.to_s+" uniq compounds in complete-set (written to data/03/#{d}_uniq.csv)"
+    else
+      raise "complete-set inchis not found in features-set #{@inchis.inspect}" unless @inchis.size==0
+      puts "inchis in features-set uniq and all included in complete-set"
+    end
+
+  end
+
+  puts ""
+
+end
+
-- 
cgit v1.2.3