Script to swap alleles in imputed .gprobs files (genotype likelihoods), e.g., from IMPUTE2

swapgeno.rb

# Script to swap alleles in imputed .gprobs files (genotype likelihoods), e.g., from IMPUTE2
# ruby swapgeno.rb genotypes.gprobs swap_snps.txt swapped.gprobs
# swap_snps.txt is a list of SNPs for which genotypes should be swapped (one per line in file)

# Read list of snps to swap
swap_snps = {}
File.open(ARGV[1]) { |sf| sf.each { |l| swap_snps[l.chomp] = true } }

File.open(ARGV[2],"w") do |outfile|
        File.open(ARGV[0]) do |file|
                file.each do |line|
                        fields =  line.chomp.split(" ")

                        next unless swap_snps[fields[1]] == true

                        i=5
                        new_fields = fields[0..2]
                        new_fields << fields[4]
                        new_fields << fields[3]

                        begin
                                rgeno = [fields[i],fields[i+1],fields[i+2]].reverse

                                # Check that we are getting the right three fields
                                sum = rgeno.inject { |i,j| i.to_f+j.to_f }
                                throw "wrong sum=#{sum}" unless (sum-1).abs < 0.01

                                1.upto(3) { |i| new_fields << rgeno.shift }
                                i=i+3
                        end while i < fields.size
                        outfile.puts new_fields.join(" ")
                end
        end
end