1 files changed, 41 insertions, 0 deletions
diff --git a/scripts/pa-groups.rb b/scripts/pa-groups.rb
new file mode 100755
index 0000000..b183116
--- /dev/null
+++ b/scripts/pa-groups.rb
@@ -0,0 +1,41 @@
+#!/usr/bin/env ruby
+group_data = File.readlines(ARGV[0]).collect{|l| l.chomp.split(",")}
+predictions = File.readlines(ARGV[1]).collect{|l| l.chomp.split(",")}
+group_names = group_data.shift
+group_names.shift
+algo_names = predictions.shift
+algo_names.shift
+groups = {}
+group_data.each do |d|
+  smi = d.shift
+  groups[smi] ||= []
+  d.each_with_index do |v,i|
+    groups[smi] << group_names[i] if v == "1"
+  end
+end
+mut = {}
+n = {}
+predictions.each do |pred|
+  smi = pred.shift
+  pred.each_with_index do |p,i|
+    algo = algo_names[i]
+    group_names.each do |g|
+      mut[g] ||= {}
+      n[g] ||= {}
+      mut[g][algo] ||= 0
+      n[g][algo] ||= 0
+      if groups[smi].include? g
+        n[g][algo]+=1
+        mut[g][algo]+=1 if p == "1"
+      end
+    end
+  end
+end
+
+q = "'"
+puts ["'PA Group'","'Model'","'Frequency'"].join(",")
+mut.each do |g,val|
+  val.each do |a,n_mut|
+    puts [q+g+q,q+a.sub("high-confidence","HC").upcase.sub(/-LR$/,"-LR-sgd").sub("LR2","LR-scikit")+q,100.0*n_mut/n[g][a]].join(",")
+  end
+end