Estimating the proportion of neutral mutations

Publisher Website
Google Scholar
Abstract
A maximum likelihood method is used for the estimation of the proportion of gene mutations that are selectively neutral. Random samples, including both neutral and deleterious alleles, are simulated by computer, and the allele frequencies are determined. The estimation method is tried out on each sample, as if the latter were for a single locus. Assuming that it is not known which alleles are neutral in the sample (as would often be the case in practice), the estimation method usually leads to the erroneous conclusion that all alleles are neutral at that locus. When it is known which alleles are neutral in the sample (and such is the case for the simulated data), the method attempts to estimate the rates of neutral and deleterious mutations and the selective disadvantage of the latter class. The estimates are somewhat biased and have relatively high variances. The results are compared, were possible, with those for an estimation method introduced by Kimura. The method is not, here, extended to samples from many independent loci. Presumably, if such samples were available and could all be assumed to be homogeneous with respect to the mutation and selection parameters, then the maximum likelihood estimation method would achieve better results.

This publication has 4 references indexed in Scilit: