mean_depths.rb

def mean_std_dev(arr,n)
	sum=0
	mean = arr.inject{|sum,x| sum + x } / n.to_f 
	total=0
	[mean, Math.sqrt(arr.inject{|total,x| total + ((mean-x)*(mean-x)) } / n.to_f) ]
end

def percentile_depths(arr,n,pct)
	samples_in_percentile = (n * pct).floor
	(arr.sort.reverse.slice(0,samples_in_percentile)).min
end

def depth_percentages(arr,n)
	more_than_n = 0
	arr.each do |depth|
		more_than_n = more_than_n + 1 if depth >= n	
	end
	more_than_n.to_f / arr.size
end

prev=nil
File.open("depths.tsv") do |file|
	file.each do |line|
		#puts line.slice(0,100)
		next if line =~ /^#.*/ 

		fields = line.chomp.split("\t")

		next if [ fields[0], fields[1] ] == prev

		depths = []
		8.upto(fields.size-1) do |i|
			depths << ((fields[i].split(/:/))[1]).to_i
		end
		depth_mean = mean_std_dev(depths,depths.size)
		puts "#{fields[0]} #{fields[1]} #{fields[2]} #{depth_mean[0]} #{depth_mean[1]} " +
			percentile_depths(depths,depths.size,0.99).inspect + " " +
			percentile_depths(depths,depths.size,0.95).inspect + " " + 
			depth_percentages(depths,20).inspect + " " + 
			depth_percentages(depths,1).inspect 

		prev = [fields[0], fields[1]]
	end
end