From 87eb7cc1e079821c2f7c5e101e7e392e9bd10f00 Mon Sep 17 00:00:00 2001
From: davor <davor@toxcreate3.in-silico.ch>
Date: Tue, 24 May 2011 09:35:11 +0200
Subject: Fixing regression detection

---
 lib/parser.rb | 66 +++++++++++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 51 insertions(+), 15 deletions(-)

diff --git a/lib/parser.rb b/lib/parser.rb
index 7bdee95..8deaa91 100644
--- a/lib/parser.rb
+++ b/lib/parser.rb
@@ -277,7 +277,23 @@ module OpenTox
       def load_spreadsheet(book)
         book.default_sheet = 0
         add_features book.row(1)
-        2.upto(book.last_row) { |i| add_values book.row(i) }
+
+        # AM: fix mixed read in
+        regression_features=false
+        2.upto(book.last_row) { |i| 
+          row = book.row(i)
+          smiles = row.shift
+          row.each_index do |i|
+            value = row[i]
+            type = feature_type(value)
+            if type == OT.NumericFeature
+              regression_features=true
+              break
+            end
+          end
+        }
+        
+        2.upto(book.last_row) { |i| add_values book.row(i),regression_features }
         warnings
         @dataset
       end
@@ -289,7 +305,23 @@ module OpenTox
         row = 0
         input = csv.split("\n")
         add_features split_row(input.shift)
-        input.each { |row| add_values split_row(row) }
+
+
+        # AM: fix mixed read in
+        regression_features=false
+        input.each { |row| 
+          row = split_row(row)
+          smiles = row.shift
+          row.each_index do |i|
+            value = row[i]
+            type = feature_type(value)
+            if type == OT.NumericFeature
+              regression_features=true
+              break
+            end
+          end
+        }
+        input.each { |row| add_values split_row(row),regression_features }
         warnings
         @dataset
       end
@@ -335,7 +367,7 @@ module OpenTox
         end
       end
 
-      def add_values(row)
+      def add_values(row, regression_features=false)
 
         smiles = row.shift
         compound = Compound.from_smiles(smiles)
@@ -353,19 +385,23 @@ module OpenTox
 
           @feature_types[feature] << type 
 
-          case type
-          when OT.NominalFeature
-            case value.to_s
-            when TRUE_REGEXP
-              val = true
-            when FALSE_REGEXP
-              val = false
-            end
-          when OT.NumericFeature
+          if (regression_features)
             val = value.to_f
-          when OT.StringFeature
-            val = value.to_s
-            @activity_errors << smiles+", "+row.join(", ")
+          else
+            case type
+            when OT.NominalFeature
+              case value.to_s
+              when TRUE_REGEXP
+                val = true
+              when FALSE_REGEXP
+                val = false
+              end
+            when OT.NumericFeature
+              val = value.to_f
+            when OT.StringFeature
+              val = value.to_s
+              @activity_errors << smiles+", "+row.join(", ")
+            end
           end
           if val!=nil
             @dataset.add(compound.uri, feature, val)
-- 
cgit v1.2.3


From 4a7ba2adb0743cd225ad5c2cf9f71c896d87b157 Mon Sep 17 00:00:00 2001
From: davor <davor@toxcreate3.in-silico.ch>
Date: Tue, 24 May 2011 10:45:53 +0200
Subject: Created dedicated function for value sweeping

---
 lib/parser.rb | 35 +++++++++++++++++------------------
 1 file changed, 17 insertions(+), 18 deletions(-)

diff --git a/lib/parser.rb b/lib/parser.rb
index 8deaa91..4984292 100644
--- a/lib/parser.rb
+++ b/lib/parser.rb
@@ -282,15 +282,8 @@ module OpenTox
         regression_features=false
         2.upto(book.last_row) { |i| 
           row = book.row(i)
-          smiles = row.shift
-          row.each_index do |i|
-            value = row[i]
-            type = feature_type(value)
-            if type == OT.NumericFeature
-              regression_features=true
-              break
-            end
-          end
+          regression_features = detect_regression_features row
+          break if regression_features=true
         }
         
         2.upto(book.last_row) { |i| add_values book.row(i),regression_features }
@@ -311,21 +304,15 @@ module OpenTox
         regression_features=false
         input.each { |row| 
           row = split_row(row)
-          smiles = row.shift
-          row.each_index do |i|
-            value = row[i]
-            type = feature_type(value)
-            if type == OT.NumericFeature
-              regression_features=true
-              break
-            end
-          end
+          regression_features = detect_regression_features row
+          break if regression_features=true
         }
         input.each { |row| add_values split_row(row),regression_features }
         warnings
         @dataset
       end
 
+
       private
 
       def warnings
@@ -367,6 +354,18 @@ module OpenTox
         end
       end
 
+      def detect_regression_features row
+        regression_features=false
+        row.each_index do |i|
+          value = row[i]
+          type = feature_type(value)
+          if type == OT.NumericFeature
+            regression_features=true
+          end
+        end
+        regression_features
+      end
+
       def add_values(row, regression_features=false)
 
         smiles = row.shift
-- 
cgit v1.2.3