From 06f93e7c8e95846bce9ed12a16d624507105ae67 Mon Sep 17 00:00:00 2001 From: rautenberg Date: Wed, 14 Sep 2016 13:23:06 +0200 Subject: add section 5 to report 'Defining the applicability domain' --- lib/report.rb | 35 ++++++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-) diff --git a/lib/report.rb b/lib/report.rb index dbb4df5..6f6cc9e 100644 --- a/lib/report.rb +++ b/lib/report.rb @@ -30,6 +30,7 @@ get "/report/:id/?" do if File.directory?("#{File.dirname(__FILE__)}/../../lazar") lazar_commit = `cd #{File.dirname(__FILE__)}/../../lazar; git rev-parse HEAD`.strip + lazar_commit = "https://github.com/opentox/lazar/tree/#{lazar_commit}" else lazar_commit = "https://github.com/opentox/lazar/releases/tag/v#{Gem.loaded_specs["lazar"].version}" end @@ -75,7 +76,7 @@ get "/report/:id/?" do report.value "algorithm_type", "#{model.class.to_s.gsub('OpenTox::Model::Lazar','')}" # Explicit algorithm 4.2 - report.change_catalog :algorithms_catalog, :algorithms_catalog_1, {:definition => "see Helma 2016 and lazar.in-silico.ch, submitted version: https://github.com/opentox/lazar/tree/#{lazar_commit}", :description => "modified k-nearest neighbor classification with activity specific similarities, weighted voting and exhaustive enumeration of fragments and neighbors"} + report.change_catalog :algorithms_catalog, :algorithms_catalog_1, {:definition => "see Helma 2016 and lazar.in-silico.ch, submitted version: #{lazar_commit}", :description => "modified k-nearest neighbor classification with activity specific similarities, weighted voting and exhaustive enumeration of fragments and neighbors"} report.ref_catalog :algorithm_explicit, :algorithms_catalog, :algorithms_catalog_1 # Descriptors in the model 4.3 @@ -95,6 +96,38 @@ get "/report/:id/?" do # Chemicals/Descriptors ratio 4.7 report.value "descriptors_chemicals_ratio", "not applicable (classification based on activities of neighbors, descriptors are used for similarity calculation)" + # Description of the applicability domain of the model 5.1 + report.value "app_domain_description", "<html> + <head> + + </head> + <body> + <p> + The applicability domain (AD) of the training set is characterized by + the confidence index of a prediction (high confidence index: close to + the applicability domain of the training set/reliable prediction, low + confidence: far from the applicability domain of the + trainingset/unreliable prediction). The confidence index considers (i) + the similarity and number of neighbors and (ii) contradictory examples + within the neighbors. A formal definition can be found in Helma 2006. + </p> + <p> + The reliability of predictions decreases gradually with increasing + distance from the applicability domain (i.e. decreasing confidence index) + </p> + </body> + </html>" + + # Method used to assess the applicability domain 5.2 + report.value "app_domain_method", "see Helma 2006 and Maunz 2008" + + # Software name and version for applicability domain assessment 5.3 + report.change_catalog :software_catalog, :software_catalog_3, {:name => "lazar, submitted version: #{lazar_commit}", :description => "integrated into main lazar algorithm", :number => "2", :url => "https://lazar.in-silico.ch", :contact => "info@in-silico.ch"} + report.ref_catalog :app_domain_software, :software_catalog, :software_catalog_3 + + # Limits of applicability 5.4 + report.value "applicability_limits", "Predictions with low confidence index, unknown substructures and neighbors that might act by different mechanisms" + # output response['Content-Type'] = "application/xml" return report.to_xml -- cgit v1.2.3