Computing probabilities of homozygosity by descent.

A person is autozygous at a locus if the person inherits the same allele twice identical by descent along two distinct paths from the same ancestor. Autozygosity is a common cause of recessive diseases in inbred populations. Homozygosity mapping uses this fact to locate the genes that cause recessive diseases. The probability of autozygosity can be used to estimate the probability of a true positive and of a false positive in homozygosity mapping. Thompson [1994] and Guo [1997] therefore studied the problem of computing the prior, unconditional (multilocus) probability of autozygosity (MPA). I consider a different quantity: the interval probability of autozygosity (IPA). The two measures are identical in the single-locus case. IPA has two notable advantages over MPA: 1. IPA does not include the possibility of heterozygous regions between the homozygous markers. 2. IPA can be computed in time that is polynomial in the pedigree size. My polynomial-time algorithm for the single-locus case solves a problem mentioned by Guo. I implemented a program to compute the IPA. The contribution of this work is the application of basic, abstract methods from computer science to address a problem in genetics.