Mutations and the probabilistic approach to incompatible paternity tests

Abstract

Paternity tests are currently carried out by typing a set of microsatellite loci (STRs). This class of polymorphisms is known to be prone to germline mutations. The gametic mutability of those STR loci that are most often used for paternity tests has been closely examined in studies [1,3] which have estimated overall mutation rates per locus, l, from a considerable volume of segregations. On the basis of these studies, there is now general consensus in defining mutations as non-sporadic events in human lineages. In the evidentiary context of a typical forensic test, mutations may lead to apparent paternal/parental exclusions. This poses problems in the interpretation of the underlying biological evidence, and the soundness of some ad hoc rules traditionally used to manage incompatible tests (such as: ‘‘non-fatherhood is certain after the occurrence of two/more than two exclusions’’) is not guaranteed. We have recently shown how a probabilistic analysis can be applied to incompatible genetic evidence in paternity tests, leading to its quantification in terms of a likelihood ratio, LR, for paternity vs. non-paternity. Our approach incorporates the following ingredients. (a) ‘‘Incompatible’’ genotype triplets explicable by assuming a single mutation are considered. (b) Arbitrary, possibly different, mutation rates l in paternal and maternal germlines are allowed. (c) For each locus, allele-specific mutation transition rates (i.e. for allele x to mutate to allele y) are derived in terms of l, based on reasonable biological assumptions including stepwise mutation and stationarity of gene frequencies [2]. (d) For each locus showing incompatibility, the probability (given the parental genotypes) of obtaining the child’s genotype under the hypothesis of paternity (assuming a