HMMfile splitter

gistfile1.rb

#!/usr/bin/env ruby

if ARGV.length != 2 then
  STDERR.puts "Usage: #{$0} <pfamfile> <directory>"
  exit 1
end

file = File.open(ARGV[0]).read

seen_accs = {}

file.split('//').each do |entry|
  acc = nil
  name = nil
  leng = nil
  desc = nil
  entry.each_line do |l|
    unless (m = /ACC\s+(.+)\./.match(l)).nil?
      acc = m[1]
    end
    unless (m = /LENG\s+(.+)/.match(l)).nil?
      leng = m[1]
    end
    unless (m = /DESC\s+(.+)/.match(l)).nil?
      desc = m[1]
    end
    unless (m = /NAME\s+(.+)/.match(l)).nil?
      name = m[1]
    end
  end
  if !(seen_accs[acc].nil?) then
    puts "acc seen twice: #{acc} (#{name})"
    exit 1
  else
    seen_accs[acc] = true
    #xxx
    puts "#{acc}\t#{name}\t#{desc}\t#{leng}"
    #xxx
  end
  File.open("#{ARGV[1]}/#{acc}_fs.hmm", "w+") do |f|
    f.write("#{entry.lstrip}")
    f.puts('//')
  end
end