The Size Distribution of Homozygous Segments in the Human Genome

Abstract

In the field of population genetics, there is a rich history of advancements made by noting discrepancies between models and theory. If observed segments of autozygosity differ in size from model predictions, we can begin to identify several possible reasons. First, natural selection will remove from the population those individuals who are autozygous for very deleterious alleles, leaving a smaller-than-expected block size. We do not tend to keep track of consanguinity if the potential common ancestor is more than four or so generations back, but having common ancestors five, 10, or even 20 generations ago is far from being “outbred” in the context of tracking homozygous segments in entire genomes. Very large segments can remain intact for long periods (fig. 1afig. 1a). There will be large sampling problems to face, and the identification of homozygous blocks will depend on marker density and heterozygosity, as Broman and Weber (1999xLong homozygous chromosomal segments in the CEPH families. Broman, KW and Weber, JL. Am J Hum Genet. 1999; 65: 1493–1500Abstract | Full Text | Full Text PDF | PubMed | Scopus (121)See all References1999 [in this issue]) illustrate. Population growth will also distort the distribution of times back to common ancestry (Bertorelle and Slatkin 1995xThe number of segregating sites in expanding human populations, with implications for estimates of demographic parameters. Bertorelle, G and Slatkin, M. Mol Biol Evol. 1995; 12: 887–892PubMedSee all References1995), and this may be especially evident in distributions of homozygous segments. Regions of the genome with low levels of recombination per megabase are expected to have larger segments of homozygosity. Observation of unexpectedly large homozygous blocks may also have a genetic cause, such as uniparental disomy (Smith et al. 1994xFamilial unbalanced translocation t(8;15)(p23.3;q11) with uniparental disomy in Angelman syndrome. Smith, A, Deng, ZM, Beran, R, Woodage, T, and Trent, RJ. Hum Genet. 1994; 93: 471–473Crossref | PubMed | Scopus (25)See all References1994; Martin et al. 1999xMaternal uniparental disomy of chromosome 14 confined to an interstitial segment (14q23-14q24.2). Martin, RA, Sabol, DW, and Rogan, PK. J Med Genet. 1999; 36: 633–636PubMedSee all References1999; Uehara et al. 1999xComplete androgen insensitivity in a 47,XXY patient with uniparental disomy for the X chromosome. Uehara, S, Tamura, M, Nata, M, Kanetake, J, Hashiyada, M, Terada, Y, Yaegashi, N et al. Am J Med Genet. 1999; 86: 107–111Crossref | PubMed | Scopus (9)See all References1999). Our ascertainment for uniparental disomy is through clinical cases, and, if some uniparental disomy for some chromosomes is asymptomatic, this phenomenon may well be more common than currently thought.Coalescence approaches will allow us to take the exciting step of turning these problems around, making inferences, and estimating population parameters from observed distributions of homozygous segment lengths. Although classical population-genetics theory tells us that one can have the same net inbreeding coefficient either with one short path of common ancestry or with several longer paths, the consequences will be very different for the distribution of blocks of homozygosity. The complete genome sequence of an individual will give the complete distribution of homozygous segments, and this distribution will allow unprecedented resolution for inferences about the number and depth of common ancestors. Many interesting inferential challenges will arise when we consider the population genetics of whole genomes.