diff options
Diffstat (limited to 'paper/create-median-correlation.rb')
-rw-r--r-- | paper/create-median-correlation.rb | 28 |
1 files changed, 28 insertions, 0 deletions
diff --git a/paper/create-median-correlation.rb b/paper/create-median-correlation.rb new file mode 100644 index 0000000..6aeee11 --- /dev/null +++ b/paper/create-median-correlation.rb @@ -0,0 +1,28 @@ +require_relative 'include.rb' + +old = Dataset.from_csv_file File.join(DATA,"mazzatorta.csv") +new = Dataset.from_csv_file File.join(DATA,"swiss.csv") + +common_compound_ids = (old.compound_ids & new.compound_ids).uniq + +data = [] +common_compound_ids.each do |cid| + c = Compound.find cid + old_values = old.values(c,old.features.first) + new_values = new.values(c,new.features.first) + identical = old_values & new_values + unless identical.empty? + old_values -= identical + new_values -= identical + end + unless old_values.empty? or new_values.empty? + data << [c.smiles,old_values.mean,new_values.mean] + end +end + +data.sort!{|a,b| a[1] <=> b[1]} + +CSV.open(File.join(DATA,"common-median.csv"),"w+") do |csv| + csv << ["SMILES","mazzatorta","swiss"] + data.each{|r| csv << r} +end |