forester/ruby/scripts/scoptastic.rb

   1 #!/usr/local/bin/ruby -w
   2 #
   3 # = scoptastic
   4 #
   5 # Copyright::  Copyright (C) 2008-2009 Christian M. Zmasek.
   6 #              All rights reserved.
   7 # License::    GNU Lesser General Public License (LGPL)
   8 #
   9 # $Id: scoptastic.rb,v 1.3 2008/08/28 17:09:07 cmzmasek Exp $
  10 #
  11 # To create Pfam id to SCOP mappings, one for each of four levels of SCOP
  12 # classification.
  13 #
  14 # Created 2008-06-25 in San Diego, CA, USA by CMZ
  15 #
  16 # Usage: scoptastic.rb scoptastic.rb <Pfam id to ac map file, e.g.
  17 # pfam_summarize.rb output> <Pfam ac to SCOP classification map file> <Pfam id
  18 # to SCOP outfile root>
  19
  20
  21 require 'iconv'
  22
  23 module ForesterScripts
  24
  25     if RUBY_VERSION !~ /1.9/
  26                       puts( "Your ruby version is #{RUBY_VERSION}, expected 1.9.x " )
  27                       exit( -1 )
  28                 end
  29
  30     CLASS_LEVEL_SUFFIX       = "_SCOP_2_CLASS"
  31     FOLD_LEVEL_SUFFIX        = "_SCOP_3_FOLD"
  32     SUPERFAMILY_LEVEL_SUFFIX = "_SCOP_4_SUPERFAMILY"
  33     FAMILY_LEVEL_SUFFIX      = "_SCOP_5_FAMILY"
  34
  35     SEP = "\t"
  36     LINE_DELIMITER  = "\n"
  37
  38     if ( ARGV == nil || ARGV.length != 3 )
  39         puts( "usage: scoptastic.rb <Pfam id to ac map file, e.g. pfam_summarize.rb output> <Pfam ac to SCOP classification map file> <Pfam id to SCOP outfile root>" )
  40         exit( -1 )
  41     end
  42
  43     pfam_id_to_ac   = ARGV[ 0 ]
  44     pfam_ac_to_scop = ARGV[ 1 ]
  45     outfile         = ARGV[ 2 ]
  46
  47     if ( !File.exists?( pfam_id_to_ac ) )
  48         puts( "Pfam id to ac map file [" + pfam_id_to_ac + "] does not exist" )
  49         exit( -1 )
  50     end
  51     if ( !File.exists?( pfam_ac_to_scop ) )
  52         puts( "Pfam ac to SCOP classification map file [" + pfam_ac_to_scop + "] does not exist" )
  53         exit( -1 )
  54     end
  55     if ( File.exists?( outfile + CLASS_LEVEL_SUFFIX ) )
  56         puts( "Outfile [" + outfile + CLASS_LEVEL_SUFFIX + "] already exists" )
  57         exit( -1 )
  58     end
  59     if ( File.exists?( outfile +  FOLD_LEVEL_SUFFIX ) )
  60         puts( "Outfile [" + outfile +  FOLD_LEVEL_SUFFIX + "] already exists" )
  61         exit( -1 )
  62     end
  63     if ( File.exists?( outfile + SUPERFAMILY_LEVEL_SUFFIX ) )
  64         puts( "Outfile [" + outfile + SUPERFAMILY_LEVEL_SUFFIX + "] already exists" )
  65         exit( -1 )
  66     end
  67     if ( File.exists?( outfile + FAMILY_LEVEL_SUFFIX ) )
  68         puts( "Outfile [" + outfile + FAMILY_LEVEL_SUFFIX + "] already exists" )
  69         exit( -1 )
  70     end
  71
  72     ic = Iconv.new( 'UTF-8//IGNORE', 'UTF-8' )
  73
  74     pfam_ac_to_id_map = Hash.new
  75
  76     pfam_ac_to_scop_map = Hash.new
  77
  78     count = 0
  79
  80     File.open( pfam_id_to_ac  ) do | file |
  81         while line = file.gets
  82             line = ic.iconv( line )
  83             if ( line !~ /^#/ && line =~ /\S/ )
  84                 if ( line =~ /^(\S+)\s+(PF\d+)/ )
  85                     pfam_ac_to_id_map[ $2 ] = $1
  86                     count += 1
  87                 else
  88                     puts( "Pfam id to ac map file [" + pfam_id_to_ac + "] format error [line: " + line + "]" )
  89                     exit( -1 )
  90                 end
  91             end
  92         end
  93     end
  94     puts()
  95     puts( "Extracted #{count} Pfam id to ac mappings from file [#{pfam_id_to_ac}]" )
  96
  97     count = 0
  98     File.open( pfam_ac_to_scop ) do | file |
  99         while line = file.gets
 100             line = ic.iconv( line )
 101             if ( line !~ /^#/ && line =~ /\S/ )
 102                 if ( line =~ /^(PF\d+)\.?\d*\s+([a-z]\.\d+\.\d+\.\d+)/ )
 103                     pfam_ac_to_scop_map[ $1 ] = $2
 104                     count += 1
 105                 else
 106                     puts( "Pfam ac to SCOP classification map file [" + pfam_ac_to_scop + "] format error [line: " + line + "]" )
 107                     exit( -1 )
 108                 end
 109             end
 110         end
 111     end
 112
 113     puts( "Extracted #{count} Pfam ac to SCOP classification mappings from file [#{pfam_ac_to_scop}]" )
 114
 115     out_class_level = File.open( outfile + CLASS_LEVEL_SUFFIX, 'w' )
 116     out_fold_level = File.open( outfile + FOLD_LEVEL_SUFFIX  , 'w' )
 117     out_superfamily_level = File.open( outfile + SUPERFAMILY_LEVEL_SUFFIX, 'w' )
 118     out_family_level = File.open( outfile + FAMILY_LEVEL_SUFFIX, 'w' )
 119
 120     count = 0
 121     pfam_ac_to_scop_map.each { | pfam_ac,scop |
 122         if ( pfam_ac_to_id_map.has_key?( pfam_ac ) )
 123             pfam_id = pfam_ac_to_id_map[ pfam_ac ]
 124             scop_split = scop.split( "\." )
 125
 126             out_class_level.write( pfam_id )
 127             out_fold_level.write( pfam_id )
 128             out_superfamily_level.write( pfam_id )
 129             out_family_level.write( pfam_id )
 130
 131             out_class_level.write( SEP )
 132             out_fold_level.write( SEP )
 133             out_superfamily_level.write( SEP )
 134             out_family_level.write( SEP )
 135
 136             out_class_level.write( scop_split[ 0 ] )
 137             out_fold_level.write( scop_split[ 0 ] + "." + scop_split[ 1 ] )
 138             out_superfamily_level.write( scop_split[ 0 ] + "." + scop_split[ 1 ] + "." + scop_split[ 2 ] )
 139             out_family_level.write( scop )
 140
 141             out_class_level.write( LINE_DELIMITER )
 142             out_fold_level.write( LINE_DELIMITER )
 143             out_superfamily_level.write( LINE_DELIMITER )
 144             out_family_level.write( LINE_DELIMITER )
 145             count += 1
 146         else
 147             puts( "Pfam ac #{pfam_ac} not found in Pfam id to ac map file [" + pfam_id_to_ac + "]" )
 148             exit( -1 )
 149         end
 150     }
 151
 152     out_class_level.close
 153     out_fold_level.close
 154     out_superfamily_level.close
 155     out_family_level.close
 156
 157     puts()
 158     puts( "Wrote #{count} Pfam id to SCOP mappings to files '#{outfile + CLASS_LEVEL_SUFFIX}', '#{outfile + FOLD_LEVEL_SUFFIX}', '#{outfile + SUPERFAMILY_LEVEL_SUFFIX}', and '#{ outfile + FAMILY_LEVEL_SUFFIX}'" )
 159     puts( "OK" )
 160     puts()
 161
 162 end # module ForesterScripts
 163