ESTIMATION OF MUTATION-RATES FROM PARENTAGE EXCLUSION DATA - APPLICATIONS TO STR AND VNTR LOCI

Nonpaternity is a common source of bias in estimating mutation rates w hen they are obtained from family data showing discordance of parental and children's genotypes. With the availability of hypervariable DNA markers, this source of bias can be largely eliminated. However, the p roportion of cases where parentage exclusion is caused by presumed mut ation(s) of parental alleles must be adjusted to obtain a valid mutati on rate estimate. The present work derives the basis of this adjustmen t factor, called the proportional bias. This proportional bias depends upon the allele frequency distribution at the locus. The maximum and minimum bounds of the proportional bias depend on the number of allele s at the locus. Using data from Caucasian populations at tandem repeat loci commonly used for parentage testing and forensic identification purposes, we show that when mutation rates are estimated at these loci , the proportional bias is generally very close to the maximum possibl e value for the observed number of alleles (or binned fragment sizes) at each locus. The expected proportional bias decreases with increasin g mutation rate at a locus. For the short tandem repeat loci, without bias correction, the direct count method can result in an underestimat ion of up to 60% of their true value. In contrast, for the minisatelli te VNTR loci, even with crude measurements on allele sizes, we show th at the absolute proportional bias is generally below the coefficient o f variation of the direct estimates.