require File.join(File.dirname(__FILE__),'../lib/rake_helper')

define_source_tasks  "h.sapiens" => "ftp://ftp.genome.jp/pub/kegg/genes/organisms/hsa/H.sapiens.ent",
  "hsa_gene_map.tab" => "ftp://ftp.genome.jp/pub/kegg/pathway/organisms/hsa/hsa_gene_map.tab",
  "drugs" => "ftp://ftp.genome.jp/pub/kegg/medicus/drug/drug",
  "pathways" => "ftp://ftp.genome.jp/pub/kegg/pathway/pathway"


file :identifiers => 'source/h.sapiens' do |t|
  pairs = {}
  entry = nil
  Open.read(t.prerequisites.first).each do |line|
    if line =~ /^ENTRY\s+(\d+)/
      entry = $1
      next
    end

    if line =~ /Ensembl: (ENSG\d+)/
      pairs[entry] = $1
    end
  end

  Open.write(t.name, ['#Ensembl Gene ID','KEGG Gene ID'] * "\t" + "\n" + pairs.collect{|entry, ens| [ens, "hsa:" + entry] * "\t"} * "\n")
end

file :gene_drug => 'source/drugs' do |t|
  pairs = {}
  drug = nil
  Open.read(t.prerequisites.first).
    scan(/^[A-Z].*?(?:^[A-Z])/sm).select{|line| line =~ /^ENTRY|TARGET/}.collect{|line| line.sub(/\s+/,' ')}.each do |line|
      if line =~ /^ENTRY\s+(\w+)/
        drug = $1
        next
      end

      if line =~ /TARGET.*?\[HSA:(.*?)\]/
        genes = $1.split(/\s/)
        genes.each do |gene|
          pairs[gene] ||= []
          pairs[gene] << drug
        end
      end
  end

  Open.write(t.name, ['#KEGG Gene ID', 'KEGG Drug ID'] * "\t" + "\n" + pairs.collect{|gene, drugs| ["hsa:" + gene, drugs * "|" ] * "\t"} * "\n")
end

file :drugs => 'source/drugs' do |t|
  info = {}
  drug = nil
  Open.read(t.prerequisites.first).
    scan(/^[A-Z].*?(?:^[A-Z])/sm).select{|line| line =~ /^ENTRY|NAME|DBLINKS/}.collect{|line| line.sub(/\s+/,' ')}.each do |line|
      if line =~ /^ENTRY\s+(\w+)/
        drug = $1
        next
      end

      if line =~ /^NAME(.*)/
        names = $1.split(/;/)
        names.each do |name|
          info[drug] ||= [[],[]]
          info[drug][0] << name.chomp.strip
        end
      end

      if line =~ /^DBLINKS(.*)/
        $1.match(/PubChem: (\d*)/)
        pubchem = $1
        next unless pubchem
        info[drug] ||= [[],[]]
        info[drug][1] << pubchem.chomp.strip
      end
  end

  Open.write(t.name, ['#KEGG Drug ID', 'KEGG Drug Name', 'PubChem Drug ID'] * "\t" + "\n" + info.collect{|drug, info| [drug, info.collect{|v| v * "|"} ].flatten * "\t"} * "\n")
end

file :pathways => 'source/pathways' do |t|
  descs = {}
  names = {}
  klass = {}
  pathway = nil
  Open.read(t.prerequisites.first).split(/\n/).each do |line|
    if line =~ /ENTRY\s+(\w+)/
      pathway = $1.strip
    end

    if line =~ /NAME (.*)/
      names[pathway] = $1.strip
    end

    if line =~ /DESCRIPTION (.*)/
      descs[pathway] = $1.strip
    end

    if line =~ /CLASS (.*)/
      klass[pathway] = $1.strip
    end
  end
  
  Open.write(t.name, "#: :type=:list\n" + ['#KEGG Pathway ID', 'Pathway Name', 'Pathway Description', 'Pathway Class'] * "\t" + "\n" + names.keys.collect{|pathway| [pathway, names[pathway], descs[pathway], klass[pathway]] * "\t"} * "\n")
end

process_tsv :gene_pathway, 'hsa_gene_map.tab',
  :sep2 => ' ' do 
  headers ['KEGG Gene ID', 'KEGG Pathway ID']
  data do |gene, pathway|
    "hsa:#{ gene }\t#{pathway.flatten.collect{|name| "hsa" + name} * "|"}"
  end
end

add_to_defaults [:pathways, :drugs, :gene_drug, :genes]