#! /usr/bin/gawk -f
# Last edited on 2008-06-15 08:19:48 by stolfi

BEGIN {
  abort = -1;
  USAGE = ( "check_translation < {INFILE}" );
  
  # Each line of the input should contain
  # a codon (with letters [ATCG]) and an aminoacid code (one
  # uppercase letter).  Checks whether the codon matches
  # the aminoacid code.
  
  # !!! Should read the aminoacid table from an external file !!!

  # Translation table:
  split("", table);
  setup_aminoacid_table(table);
}

(abort >= 0) { exit abort; }

# Ignore blank lines:
/^ *$/ { next; }

// {
  codon = toupper($1); amino = $2;
  if (NF != 2)
    { data_error(("wrong number of fields")); }
  if (codon !~ /^[ATCG][ATCG][ATCG]$/)
    { data_error(("bad codon format")); }
  if (amino !~ /^[A-Z]$/)
    { data_error(("bad aminoacid format")); }
  if (amino != table[codon])
    { data_error(("incorrect translation for \"" codon "\" = \"" amino "\" should be \"" table[codon] "\"")); }
}

function setup_aminoacid_table(tb )
{
  # Stores in {tb} the codon-to-aminoacid table.
  # Aminoacids are represented by capital letter.
  # The stop codons are mapped to "$".

  tb["AAA"] = "K";  
  tb["AAT"] = "N";  
  tb["AAC"] = "N";  
  tb["AAG"] = "K";  
  tb["ATA"] = "I";  
  tb["ATT"] = "I";  
  tb["ATC"] = "I";  
  tb["ATG"] = "M";  
  tb["ACA"] = "T";  
  tb["ACT"] = "T";  
  tb["ACC"] = "T";  
  tb["ACG"] = "T";  
  tb["AGA"] = "R";  
  tb["AGT"] = "S";  
  tb["AGC"] = "S";  
  tb["AGG"] = "R";  
  tb["TAA"] = "$";  
  tb["TAT"] = "Y";  
  tb["TAC"] = "Y";  
  tb["TAG"] = "$";  
  tb["TTA"] = "L";  
  tb["TTT"] = "F";  
  tb["TTC"] = "F";  
  tb["TTG"] = "L";  
  tb["TCA"] = "S";  
  tb["TCT"] = "S";  
  tb["TCC"] = "S";  
  tb["TCG"] = "S";  
  tb["TGA"] = "$";  
  tb["TGT"] = "C";  
  tb["TGC"] = "C";  
  tb["TGG"] = "W";  
  tb["CAA"] = "Q";  
  tb["CAT"] = "H";  
  tb["CAC"] = "H";  
  tb["CAG"] = "Q";  
  tb["CTA"] = "L";  
  tb["CTT"] = "L";  
  tb["CTC"] = "L";  
  tb["CTG"] = "L";  
  tb["CCA"] = "P";  
  tb["CCT"] = "P";  
  tb["CCC"] = "P";  
  tb["CCG"] = "P";  
  tb["CGA"] = "R";  
  tb["CGT"] = "R";  
  tb["CGC"] = "R";  
  tb["CGG"] = "R";  
  tb["GAA"] = "E";  
  tb["GAT"] = "D";  
  tb["GAC"] = "D";  
  tb["GAG"] = "E";  
  tb["GTA"] = "V";  
  tb["GTT"] = "V";  
  tb["GTC"] = "V";  
  tb["GTG"] = "V";  
  tb["GCA"] = "A";  
  tb["GCT"] = "A";  
  tb["GCC"] = "A";  
  tb["GCG"] = "A";  
  tb["GGA"] = "G";  
  tb["GGT"] = "G";  
  tb["GGC"] = "G";  
  tb["GGG"] = "G";  
}

function data_error(msg)
{
  printf "  %s\n", $0 > "/dev/stderr";
  printf "** %s:%s: %s\n", FILENAME, FNR, msg > "/dev/stderr";
  abort = 1;
  exit abort;
}

function arg_error(msg)
{
  printf "** %s\n", msg > "/dev/stderr";
  printf "usage: %s\n", USAGE > "/dev/stderr";
  abort = 1;
  exit abort;
}

END {
  if (abort >= 0) { exit abort; }
}