GenePoker

README.md

Original rules by Сергей Слюсарев.

IUPAC notation

gene_poker.rb

module GenePoker
  RnaAlphabet = %w(G A U C)
  IupacComprehensions = {
    ' ' => '|',
    'N' => '.',
    'R' => '[AG]',
    'Y' => '[UC]',
    'H' => '[^G]',
  }
  AminoCodes = {
    "Ala" => "GCN",
    "Leu" => "UUR CUN",
    "Arg" => "CGN AGR",
    "Lys" => "AAR",
    "Asn" => "AAY",
    "Met" => "AUG",
    "Asp" => "GAY",
    "Phe" => "UUY",
    "Cys" => "UGY",
    "Pro" => "CCN",
    "Gln" => "CAR",
    "Ser" => "UCN AGY",
    "Glu" => "GAR",
    "Thr" => "ACN",
    "Gly" => "GGN",
    "Trp" => "UGG",
    "His" => "CAY",
    "Tyr" => "UAY",
    "Ile" => "AUH",
    "Val" => "GUN"
  }
  IupacCodesRegex = Regexp.union IupacComprehensions.keys

  def self.iupac_to_regex iupac
    Regexp.new iupac.gsub(IupacCodesRegex, IupacComprehensions)
  end

  AminoCodes.each do |acid,codes|
    AminoCodes[acid] = iupac_to_regex(codes)
  end

  def self.hands
    RnaAlphabet.repeated_combination(4)
  end

  def self.aminoacid code
    acid, _ = AminoCodes.find{|acid,regex| regex === code}
    acid
  end

  def self.produce_aminoacids hand
    hand.permutation(3).map{|code|aminoacid(code.join)}.compact.uniq
  end
end

GenePoker.hands.map do |hand|
  aminoacids = GenePoker.produce_aminoacids(hand)
  {
    hand: hand.join,
    value: aminoacids.count,
    aminoacids: aminoacids.join(' ')
  }
end.sort_by do |row|
  -row[:value]
end.each do |row|
  puts row.values.join "\t"
end

gistfile1.txt

GAUC	13	Asp Val Ala Ser Met Ile Thr Cys Tyr Arg Gln His Leu
GAAU	9	Glu Asp Val Arg Ser Lys Asn Met Ile
GAAC	9	Glu Asp Ala Arg Ser Lys Asn Thr Gln
GACC	8	Asp Ala Ser Thr Arg Gln His Pro
GAUU	8	Asp Val Ser Met Ile Cys Tyr Leu
GGAU	8	Gly Glu Asp Val Arg Ser Met Trp
GGAC	8	Gly Glu Asp Ala Arg Ser Thr Gln
AAUC	8	Asn Ile Thr Tyr Ser Gln His Leu
GGUC	8	Gly Val Ala Trp Cys Ser Arg Leu
GUUC	7	Val Ala Cys Leu Phe Ser Arg
AUUC	7	Ile Thr Tyr Leu Phe Ser His
AUCC	7	Ile Thr Tyr Ser His Leu Pro
GUCC	7	Val Ala Cys Ser Arg Leu Pro
AACC	5	Asn Thr Gln His Pro
GGUU	5	Gly Val Trp Cys Leu
GUUU	4	Val Cys Leu Phe
GGCC	4	Gly Ala Arg Pro
AUUU	4	Ile Tyr Leu Phe
GGAA	4	Gly Glu Arg Lys
UUCC	4	Phe Ser Leu Pro
AAUU	4	Asn Ile Tyr Leu
AAAC	4	Lys Asn Thr Gln
GAAA	3	Glu Arg Lys
AAAU	3	Lys Asn Ile
GGGA	3	Gly Glu Arg
GCCC	3	Ala Arg Pro
UCCC	3	Ser Leu Pro
GGGU	3	Gly Val Trp
GGGC	3	Gly Ala Arg
ACCC	3	Thr His Pro
UUUC	3	Phe Ser Leu
GGGG	1	Gly
AAAA	1	Lys
UUUU	1	Phe
CCCC	1	Pro