forester/ruby/scripts/hmm_split.rb

   1 #!/usr/local/bin/ruby -w
   2 #
   3 # = hmm_split
   4 #
   5 # Copyright::  Copyright (C) 2006-2008 Christian M. Zmasek
   6 # License::    GNU Lesser General Public License (LGPL)
   7 #
   8 # $Id: hmm_split.rb,v 1.5 2008/11/17 22:32:43 cmzmasek Exp $
   9 #
  10 # To split a Pfam HMM file into one file for each HMM.
  11 #
  12
  13
  14 module ForesterScripts
  15
  16     if RUBY_VERSION !~ /1.9/
  17                       puts( "Your ruby version is #{RUBY_VERSION}, expected 1.9.x " )
  18                       exit( -1 )
  19                 end
  20
  21
  22     if ( ARGV == nil || ARGV.length != 3 )
  23         puts( "usage: hmm_split.rb <Pfam HMM file> <outfile suffix> <outdir>" )
  24         exit( -1 )
  25     end
  26
  27        hmmfile = ARGV[ 0 ]
  28        suffix  = ARGV[ 1 ]
  29        outdir  = ARGV[ 2 ]
  30
  31        if ( !File.exists?( outdir ) )
  32            puts( "outdir [" + outdir + "] does not exist" )
  33            exit( -1 )
  34        end
  35        if ( !File.exists?( hmmfile ) )
  36            puts( "Pfam HMM file [" + hmmfile + "] does not exist" )
  37            exit( -1 )
  38        end
  39
  40        data = String.new
  41        name = String.new
  42        line_count = 0
  43        count = 0
  44
  45        File.open( hmmfile ) do | file |
  46            while line = file.gets
  47                data = data + line
  48                line_count += 1
  49                if ( line =~ /NAME\s+(.+)/ )
  50                    if name.length > 0
  51                        puts( "Pfam HMM file [" + hmmfile + "] format error [line: " + line + "]" )
  52                        exit( -1 )
  53                    end
  54                    name = $1
  55                elsif ( line =~ /\/\// )
  56                    if name.length < 1
  57                        puts( "Pfam HMM file [" + hmmfile + "] format error [line: " + line + "]" )
  58                        exit( -1 )
  59                    end
  60
  61                    outfile = outdir + '/' + name + suffix
  62                    if ( File.exists?( outfile ) )
  63                        puts( "file [" + outfile + "] already exists" )
  64                        exit( -1 )
  65                    end
  66                    open( outfile, 'w' ) do | out |
  67                        out.write( data )
  68                    end
  69                    count += 1
  70                    puts( count.to_s + ": " + name )
  71                    data = String.new
  72                    name = String.new
  73                end
  74            end
  75        end
  76
  77        puts()
  78        puts( "wrote " + count.to_s + " individual HMM files to " + outdir )
  79
  80 end