Pedigree-based relationship inference from complex DNA mixtures.

We present a general method for analysing DNA mixtures involving relatives that accounts for dropout and drop-in, mutations, silent alleles and population substructure. Whether the aim is to identify the contributors to a mixture who may be related, or to determine the relationship between individuals based on a DNA mixture, both types of problems can be handled by the method and software presented here. We focus on the latter scenario, motivated by non-invasive prenatal paternity testing where the profile of the child is available only in the form of a mixture with the mother's profile. Relationships are represented by pedigrees and can include kinship between more than two individuals. The software is freely available as a graphical user interface in the R package relMix.