diff options
author | davor <davor@toxcreate3.in-silico.ch> | 2011-05-24 09:35:11 +0200 |
---|---|---|
committer | davor <davor@toxcreate3.in-silico.ch> | 2011-05-24 09:35:11 +0200 |
commit | 87eb7cc1e079821c2f7c5e101e7e392e9bd10f00 (patch) | |
tree | a821a7ac2023026498f96d5b40df0b72fb4650bc | |
parent | 5567a6fef6fc134f484809a137525cd27fb5dfbb (diff) |
Fixing regression detection
-rw-r--r-- | lib/parser.rb | 66 |
1 files 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) |