Fig. 3
From: A covert authentication and security solution for GMOs
Correspondence between the binary text triplets B i and DNA codons C i for the specific example where n 1=A,n 2=C,n 3=G,n 4=T. Each of the text triplets is distributed over the associated codon triplets so that the resulting representation resembles the codon bias of the host genome. To demonstrate the watermarking protocol, assume there are 44 occurrences of 000 in the binary text and that the codon frequency values as determined from the entire codon frequency distribution, are: A A A,3.3 %, A A C,2.1 %, A C A,0.8 %, A C C,2.3 %, C A A,1.5 %, C A C,0.9 %, C C A,0.8 %, C C C,0.6 %, covering a total of 12.5 % of the total codon distribution. Among the codons assigned to 000, there are 100·3.3/12.5=26.6 % for AAA, 17.1 % for AAC etc. Consequently, we assign (26.6·44)/100∼12 occurrences of 000 to AAA, 8 to AAC, 3 to ACA, 8 to ACC, 5 to CAA, 3 to CAC, 3 to CCA, and 2 to CCC, covering the total 44 occurrences of 000