gistfile1.txt

#!/usr/bin/env ruby
# Christian Theil Have, Sep. 2009.

require 'rubygems'
require 'xmlsimple'

class String
  def to_prolog
    "'" + self.gsub("'",'\\\\\'') + "'"
  end

  def to_prolog_array
    arr = []
    self.downcase.each_byte { |b| arr << b.chr }
    arr.to_prolog
  end
  
  def to_prolog_atom
    self.downcase
  end
end

class Array
  def to_prolog
    to_prolog_list
  end
  
  def to_prolog_facts
    (self.map{ |e| e.to_prolog}).join('\n')
  end
  
  def to_prolog_list
    "[" + (self.map { |e| e.to_prolog}).join(',') + "]"
  end
end

class BlastMatch
  def initialize(hit_id, hit_features)
    @hit_id, @features = hit_id, hit_features
  end
  
  def to_prolog
    prolog_str = ""
    @features.each do |key,val|
        prolog_str << "blast_match(#{@hit_id.to_prolog}, #{key.to_s}, #{val})." << "\n"
    end
    prolog_str
  end
end

class BlastXML2Prolog
  def initialize(infile,outfile)
    @xml = XmlSimple.xml_in infile
    iteration = @xml['BlastOutput_iterations'][0]['Iteration'][0]
    hits = []
    iteration_hits = iteration['Iteration_hits'][0]['Hit']
    iteration_hits.each do |h|
      first_hsp = h['Hit_hsps'][0]['Hsp'][0]
      hits << BlastMatch.new(
        h['Hit_id'].first.to_s,
        {
        :hit_num => h['Hit_num'][0].to_s.to_prolog,
        :organism => h['Hit_def'][0].to_prolog,
        :evalue => first_hsp['Hsp_evalue'][0].to_s.to_prolog,
        :qseq => first_hsp['Hsp_qseq'][0].to_s.to_prolog_array,
        :hseq => first_hsp['Hsp_hseq'][0].to_s.to_prolog_array
        }
      )
    end
    File.open(outfile, "w") { |f| hits.each { |h| f << h.to_prolog } }
  end
end

if  ARGV.length != 2
  puts "Usage is blastxml2prolog inputfile outputfile"
  exit -1
end

BlastXML2Prolog.new(ARGV[0],ARGV[1])