Statistical Tests of the Coalescent Model Based on the Haplotype Frequency Distribution and the Number of Segregating Sites

Abstract

Several tests of neutral evolution employ the observed number of segregating sites and properties of the haplotype frequency distribution as summary statistics and use simulations to obtain rejection probabilities. Here we develop a "haplotype configuration test" of neutrality (HCT) based on the full haplotype frequency distribution. To enable exact computation of rejection probabilities for small samples, we derive a recursion under the standard coalescent model for the joint distribution of the haplotype frequencies and the number of segregating sites. For larger samples, we consider simulation-based approaches. The utility of the HCT is demonstrated in simulations of alternative models and in application to data from Drosophila melanogaster.S ELECTIVELY neutral models of within-species evo- (Hudson et al. 1994). Alternatively, because haplotype frequency distributions may differ greatly across demolution consist of a model that describes the genealgraphic scenarios (Nei et al. 1975;Donnelly 1996), ogy of sampled DNA sequences and a model for the haplotype tests can also help to identify deviations from stochastic process of mutation along the branches of the the demographic assumptions of the standard neutral genealogy. Typical neutral models use the coalescent model. process (Nordborg 2001, for example) to describe the Some of the first haplotype tests, such as the Ewensgenealogy and the infinitely many sites model (WattWatterson homozygosity test, were based on the Ewens erson, 1975) for the mutation process. Many theoretical (1972) sampling theory for the infinitely many alleles predictions have been made under the standard neutral mutation model. Because "allele" in this model and model, in which the particular coalescent model chosen "haplotype" in the infinitely many sites model have the is the one with constant population size. same meaning, the Ewens (1972) theory provides the As an alternative to computationally intensive compariconditional distribution of the haplotype frequency vecsons of likelihoods of DNA sequence data under null and tor C given the sample size n and the number of distinct alternative models (Griffiths and Tavaré 1994; Kuhner haplotypes K, under the standard neutral model (Tavaré et al.

Signatures of strong population differentiation shape extended haplotypes across the human CD28 , CTLA4 , and ICOS costimulatory genes

“…A recent study systematically interrogated a number of SNPs in the CTLA4 region in the same CEPH-HGD panel (15), leading to the description of 10 haplotypes, which agree well with the ones we identified. In almost all populations, the distribution of haplotypes includes one dominant representative that accounts for approximately half of the chromosomes, and one or a few minor ones that complete the pool, as commonly observed in other genes and expected from theoretical considerations (16). Differential SNP distributions translate here as a distribution of haplotypes that is markedly different between the various population groups, with delineations that correspond precisely to geographical and historical boundaries.…”

Methods for Human Demographic Inference Using Haplotype Patterns From Genomewide Single-Nucleotide Polymorphism Data

“…Consequently, we deliberately did not use the number of SNPs in constructing the HCN statistic. To analyze full-resequencing data in a haplotype framework, a more powerful approach would also make use of information about the number of SNPs in each window (Innan et al 2005). The HCN statistic can be modified to include this information, suggesting that haplotype patterns based on full-resequencing data will be even more informative than described here.…”

“…We chose to use the number of haplotypes as a summary statistic because it is a sufficient statistic for the population mutation rate ðuÞ in the infinite-alleles model (Ewens 1972) and has been shown by simulation to be informative about population history (Depaulis and Veuille 1998;Innan et al 2005). The count of the most common haplotype was also suggested as a test statistic in the infinite-alleles model (Ewens 1973) and has been found to be correlated with haplotype homozygosity (Zeng et al 2007 and data not shown).…”

“…Many recent studies have assumed a demographic model (often the standard neutral model) and then used either LD or haplotype patterns to estimate recombination rates (Wall 2000a;Hudson 2001;Li and Stephens 2003;McVean et al 2004;Myers et al 2005). Other studies have taken the opposite approach and assumed that the recombination rate is known and then used LD or haplotype patterns to estimate demographic parameters (Reich et al 2001;Innan et al 2005;Voight et al 2005;Schaffner et al 2005;Leblois and Slatkin 2007;Tenesa et al 2007). The way in which haplotype information has been used for demographic inference is quite variable among studies.…”

Haplotype Trees and Modern Human Origins

“…The infinite-sites model is sometimes mistakenly equated with the infinite-alleles model (Innan et al, 2005). If the infinite-sites model is true, then all mutations must result in new haplotypes, so the infinite-alleles model is also true.…”