From 7a6d28d9ad32198af1feb848352731ab2fd7e2f1 Mon Sep 17 00:00:00 2001
From: Christoph Helma <helma@in-silico.ch>
Date: Fri, 25 Jun 2021 11:29:49 +0200
Subject: first revision

---
 scripts/sa37.rb | 35 +++++++++++++++++++++++++++++++++++
 1 file changed, 35 insertions(+)
 create mode 100755 scripts/sa37.rb

(limited to 'scripts/sa37.rb')

diff --git a/scripts/sa37.rb b/scripts/sa37.rb
new file mode 100755
index 0000000..e628f51
--- /dev/null
+++ b/scripts/sa37.rb
@@ -0,0 +1,35 @@
+#!/usr/bin/env ruby
+group_data = File.readlines(ARGV[0]).collect{|l| l.chomp.split(",")}
+sa37 = File.readlines(ARGV[1]).collect{|l| l.chomp.split(",")}
+group_names = group_data.shift
+group_names.shift
+sa37.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 = {}
+sa37.each do |pred|
+  smi = pred.shift
+  group_names.each do |g|
+    mut[g] ||= 0
+    n[g] ||= 0
+    if groups[smi].include? g
+      n[g]+=1
+      mut[g]+=1 if pred[0] == "1"
+    end
+  end
+end
+
+perc = {}
+mut.each do |g,m|
+  perc[g] = (100.0*m/n[g]).round
+end
+p perc
-- 
cgit v1.2.3