indel.rb

#/usr/bin/env ruby


def open_indel_file(filename)
  h = []
  open(filename) { |f|
    f.each_line do |line|
      a = line.chomp.split("\t")
      h << a
    end
  }
  h.reverse!{ |a,b| a[0].to_i <=> b[0].to_i}
  h
end

def open_fasta_file(filename)
  sequence = ""
  seq_name = ""

  open(filename){ |f|
    f.each_line do |line|
      if line =~ />(.+)$/
        seq_name = $1
      else
        sequence += line.chomp
      end
    end
  }
  return sequence, seq_name
end


if __FILE__ == $PROGRAM_NAME
  indels = open_indel_file(ARGV[0])
  sequence, seq_name = open_fasta_file(ARGV[1])
  output_file_name = "output.fasta"               # 出力ファイル名
  output_seq_length = 60                          # 出力するFASTAの1行あたりの塩基長

  indels.each do |i|
    s = i[0].to_i
    n = i[1]
    flag = i[2]
    
    if  flag == "insertion"
      sequence.insert(s, n)
    elsif flag  == "deletion"
      sequence[s, n.length] = ""
    else
      puts "unknown format : #{flag}"
    end
  end

  open(output_file_name, "w"){ |f|
    f.puts ">" + seq_name
    sequence.each_char.each_slice(output_seq_length){|l| f.puts l.join}  
    f.puts "\n"
  }
  
end

http://qa.lifesciencedb.jp/questions/328