endpoint counts

1 files changed, 51 insertions, 3 deletions

diff --git a/import.rb b/import.rb
index b086a8c..2a83b94 100644
--- a/import.rb
+++ b/import.rb
@@ -1,5 +1,6 @@
 require 'json'
 require 'yaml'
+#require_relative "../lazar/lib/lazar.rb"
 require_relative "lib/nano-lazar.rb"
 include OpenTox
 
@@ -13,7 +14,6 @@ names = []
     if substance["composition"]
       nr_cores = substance["composition"].select{|c| c["relation"] == "HAS_CORE"}.size
       puts "#{substance["compound"]["URI"]} has #{nr_cores} cores" if nr_cores !=1
-      #component = substance["composition"]#["component"]
       substance["composition"].each do |composition|
         component = composition["component"]
         if component
@@ -61,12 +61,60 @@ names = []
         #p k,v
       end
     end
+    nm.tox.uniq! if nm.tox
+    nm.p_chem.uniq! if nm.p_chem
     nanomaterials << nm
   end
 end
+
 puts "Total imported: #{nanomaterials.size}"
-puts "With TOX data: #{nanomaterials.select{|n| n.tox}.size}"
 puts "With nanoparticle characterisation: #{nanomaterials.select{|n| n.p_chem}.size}"
-puts "With TOX data and particle characterisation: #{nanomaterials.select{|n| n.tox and n.p_chem}.size}"
+modelling_data = nanomaterials.select{|n| n.tox and n.p_chem}
+puts "With TOX data: #{nanomaterials.select{|n| n.tox}.size}"
+puts "With TOX data and particle characterisation: #{modelling_data.size}"
+endpoints = modelling_data.collect{|n| n.tox.collect{|t| t.keys}}.flatten.compact.uniq
+puts "Endpoints: #{endpoints.size}"
+
+single_value_endpoints = []
+endpoint_values = {}
+
+endpoints.each do |e|
+  #json = `curl -H "Accept:application/json" "#{e}" 2>/dev/null`
+  #f = JSON.parse(json)["feature"]
+  #p k unless f.keys.size == 1
+  #k = f.keys.first
+  #p e
+  #p modelling_data.select{|n| n.tox.select{|t| t[e]}}.size
+  i = 0
+  values = []
+  modelling_data.each do |n|
+    n.tox.each do |t|
+      if t[e]
+        i += 1
+        values << t[e]
+      end
+    end
+  end
+  single_value_endpoints << e if values.uniq.size == 1
+  endpoint_values[e] = values.size unless values.uniq.size == 1
+  #puts "#{f[k]['title']} [#{f[k]['units']}]: #{i} #{values}"
+end
+
+endpoints -= single_value_endpoints
+puts "Endpoints with more than one measurement value: #{endpoints.size}"
+#endpoint_values.sort{|a,b| b[1] <=> a[1]}
+endpoint_values.select!{|k,v| v > 10}
+puts "Endpoints with more than 10 measurements: #{endpoint_values.size}"
+endpoints = endpoint_values.keys
+#puts endpoints.to_yaml
+endpoint_values.sort{|a,b| b[1] <=> a[1]}.each do |e,v|
+  json = `curl -H "Accept:application/json" "#{e}" 2>/dev/null`
+  f = JSON.parse(json)["feature"]
+  p k unless f.keys.size == 1
+  k = f.keys.first
+  p e
+  puts "#{f[k]['title']} [#{f[k]['units']}]: #{v} "
+end
+#puts "Endpoints with more than one value single_value_endpoints.size
 #puts names.sort.uniq.to_yaml
 #p nanomaterials.collect{|n| n.uri}.uniq.size