A correspondence between open reading frames in sense and antisense st
rands is expected from the hypothesis that the prototypic tripler code
was of general form RNY, where R is a purine base, N is any base, and
Y is a pyrimidine. A deficit of stop codons in the antisense strand (
and thus long open reading frames) is predicted for organisms with hig
h G + C percentages; however, two bacteria (Azotobacter vinelandii, Rh
odobacter capsulatum) have larger average antisense strand open readin
g frames than predicted from (G + C)%. The similar codon frequencies f
ound in sense and antisense strands can be attributed to the wide dist
ribution of inverted repeats (stem-loop potential) in natural DNA sequ
ences.