Confidence interval for the number of selectively neutral amino acid polymorphisms.

A statistical approach to the analysis of DNA sequences has been developed, which provides a confidence interval estimate for the proportion of naturally occurring amino acid polymorphisms that are selectively neutral. When applied to the gnd gene coding for 6-phosphogluconate dehydrogenase in a sample of seven natural isolates of Escherichia coli, the method indicates that the proportion of observed amino acid polymorphisms that are selectively neutral is unlikely to be greater than 49% (upper 95% confidence limit). On the other hand, the observations are also consistent with a model in which all of the observed amino acid substitutions are mildly deleterious with an average selection coefficient approximating 1.6 X 10(-7). Various models for the distribution of configurations at silent sites are also investigated.