Biases and balances within the genetic code

Encoding sequences show distinct nucleotide compositions within the individual codon positions, termed the position-specific nucleotide frequencies. The increased amounts of sequence data for human and E. coli are evaluated for the position-specific composition of all four nucleotides as well as the subgroups S and R. A, G and S reveal a pronounced position-specific pattern. These nucleotide biases are partially known; the characteristic A pattern was not reported before. Biases in position 31 and 2 are predominanlty caused by amino acid composition of proteins. The biases in third positions are proposed to be driven by the tendency of compensation leading to statistical gross nucleotide content of coding sequences. Possible applications of position-specific frequencies are use as a phylogenetic marker or as a basis of coding measures. (C) 1996 Academic Press, Inc.