forester/ruby/evoruby/exe/run_phylo_pipeline_x.rb

   1 #!/usr/local/bin/ruby -w
   2 #
   3 # = run_phylo_pipeline
   4 #
   5 # Copyright::  Copyright (C) 2010 Christian M. Zmasek
   6 # License::    GNU Lesser General Public License (LGPL)
   7 #
   8 # $Id Exp $
   9 #
  10 #
  11
  12 require 'fileutils'
  13
  14 module Evoruby
  15
  16   class RunPhyloPipeline
  17
  18     PFAM      = "/home/czmasek/DATA/PFAM/PFAM270X/"
  19     HMMSCAN  = "/home/czmasek/SOFTWARE/HMMER/hmmer-3.0/src/hmmscan"
  20     HSP       = "/home/czmasek/SOFTWARE/FORESTER/DEV/forester/forester/ruby/evoruby/exe/hsp.rb"
  21     D2F       = "/home/czmasek/SOFTWARE/FORESTER/DEV/forester/forester/ruby/evoruby/exe/d2f.rb"
  22     DSX       = "/home/czmasek/SOFTWARE/FORESTER/DEV/forester/forester/ruby/evoruby/exe/dsx.rb"
  23     TAP       = "/home/czmasek/SOFTWARE/FORESTER/DEV/forester/forester/ruby/evoruby/exe/tap.rb"
  24
  25     def run
  26       unless ARGV.length >= 2 && ARGV.length <= 4
  27         error "arguments are:  <min-length> " +
  28          "<neg E-value exponent for domain extraction> [E-value for hmmscan, default is 10] [hmmscan option, default is --nobias, --max for no heuristics]"
  29       end
  30
  31       length      = ARGV[ 0 ].to_i
  32       e_value_exp = ARGV[ 1 ].to_i
  33
  34       e_for_hmmscan = 10
  35       hmmscan_option = "--nobias"
  36
  37       if ARGV.length == 4
  38         hmmscan_option = ARGV[ 3 ]
  39       end
  40       if ARGV.length == 3 || ARGV.length == 4
  41         e_for_hmmscan = ARGV[ 2 ].to_i
  42       end
  43
  44       if e_value_exp < 0
  45         error "E-value exponent for domain extraction cannot be negative"
  46       end
  47       if length <= 1
  48         error "length cannot be smaller than or equal to 1"
  49       end
  50       if e_for_hmmscan < 1
  51         error "E-value for hmmscan cannot be smaller than 1"
  52       end
  53
  54       input_files = Dir.entries(".").select { |f| !File.directory?( f ) && f.downcase.end_with?( ".fasta" ) }
  55
  56       puts "Input files:"
  57       input_files.each do | input |
  58         puts input
  59       end
  60       puts
  61
  62       counter = 1
  63       input_files.each do | input |
  64
  65         puts counter.to_s + "/" +  input_files.size.to_s + " " + input + ": "
  66
  67         counter += 1
  68
  69         hmm_name = ""
  70         id_norm = false
  71         orig_input = input
  72
  73         if input.downcase.end_with?( "_ni.fasta" )
  74           hmm_name = input[ 0 .. input.length - 10 ]
  75         elsif input.downcase.end_with?( ".fasta" )
  76           id_norm = true
  77           hmm_name = input[ 0 .. input.length - 7 ]
  78           puts
  79           puts "a. identifier normalization:"
  80           cmd = "#{TAP} #{input} #{hmm_name}_ni.fasta #{hmm_name}.nim"
  81           run_command( cmd )
  82           input = hmm_name + "_ni.fasta"
  83         else
  84           error "illegal name: " + input
  85         end
  86
  87         Dir.mkdir( hmm_name )
  88
  89         puts
  90         puts "b. hmmscan:"
  91         cmd = "#{HMMSCAN} #{hmmscan_option} --domtblout #{hmm_name}/#{hmm_name}_hmmscan_#{e_for_hmmscan.to_s} -E #{e_for_hmmscan.to_s} #{PFAM}Pfam-A.hmm #{input}"
  92         run_command( cmd )
  93         puts
  94
  95         puts "c. hmmscan to simple domain table:"
  96         cmd = "#{HSP} #{hmm_name}/#{hmm_name}_hmmscan_#{e_for_hmmscan.to_s} #{hmm_name}/#{hmm_name}_hmmscan_#{e_for_hmmscan.to_s}_domain_table"
  97         run_command( cmd )
  98         puts
  99
 100         puts "d. domain table to forester format:"
 101         cmd = "#{D2F} -e=10 #{hmm_name}/#{hmm_name}_hmmscan_#{e_for_hmmscan.to_s}_domain_table #{input} #{hmm_name}/#{hmm_name}_hmmscan_#{e_for_hmmscan.to_s}.dff"
 102         run_command( cmd )
 103         puts
 104
 105         puts "e. dsx:"
 106         cmd = "#{DSX} -d -e=1e-#{e_value_exp.to_s} -l=#{length} #{hmm_name} #{hmm_name}/#{hmm_name}_hmmscan_#{e_for_hmmscan.to_s} #{input} #{hmm_name}/#{hmm_name}__#{hmm_name}__ee#{e_value_exp.to_s}_#{length}"
 107         run_command( cmd )
 108         puts
 109
 110         if id_norm
 111           FileUtils.mv "#{hmm_name}_ni.fasta", "#{hmm_name}/#{hmm_name}_ni.fasta"
 112           FileUtils.mv "#{hmm_name}.nim", "#{hmm_name}/#{hmm_name}.nim"
 113           FileUtils.cp orig_input, "#{hmm_name}/#{orig_input}"
 114         end
 115
 116         Dir.mkdir( hmm_name + "/msa" )
 117
 118         FileUtils.cp "#{hmm_name}/#{hmm_name}__#{hmm_name}__ee#{e_value_exp.to_s}_#{length}.fasta", "#{hmm_name}/msa/#{hmm_name}__#{hmm_name}__ee#{e_value_exp.to_s}_#{length}"
 119
 120       end
 121
 122     end
 123
 124     def run_command cmd
 125       puts cmd
 126       `#{cmd}`
 127     end
 128
 129     def get_base_name n
 130       if n.downcase.end_with?( "_ni.fasta" )
 131         n[ 0 .. n.length - 10 ]
 132       elsif n.downcase.end_with?( ".fasta" )
 133         n[ 0 .. n.length - 7 ]
 134       else
 135         error "illegal name: " + n
 136       end
 137     end
 138
 139     def error msg
 140       puts
 141       puts msg
 142       puts
 143       exit
 144     end
 145
 146   end
 147
 148   p = RunPhyloPipeline.new()
 149
 150   p.run()
 151
 152 end
 153
 154
 155