Figure 8

Amino acid context signatures detect genetic code alterations. In order to determine whether genetic code alterations could originate a specific triplet signature, the frequencies of amino acid contexts having leucine or serine in the middle position (ex. LYS-LEU-ASP/LYS-SER-ASP) were subtracted. Whenever this difference was higher than 0.0005 or lower than -0.0005 the respective context was considered biased towards leucine or serine, respectively. These biased neighborhoods were checked for Leu/Ser-CTG, Leu-CTA and Ser-TCA codons. The expected values were calculated for all the contexts and subtracted from the observed values. In order to normalize the bias with the total pool size for each codon-context each difference was divided by the expected value [(Obs-Exp)/Exp]. The sum of the quotients of all leucine-preferred (yellow bars) and serine-preferred (red bars) neighborhoods for each ORFeome showed the global effect. As expected, leucine CTG codons (panel-A) were more frequent in leucine-preferred contexts (yellow bars) than in serine-preferred ones (red bars). However, this signature was broken in C. albicans and D. hansenii, (where CTGs are decoded as serine and not leucine) since CTGs were associated with serine- rather than leucine-preferred neighbors. This trend was not detected in any other leucine or serine codon (Ex: Leu-CTA and Ser-TCA, panels-B and -C, respectively), indicating that genetic code alterations can be detected through codon-triplet context analysis.