Bayesian Analysis of Haplotypes for Linkage Disequilibrium Mapping

Abstract

Haplotype analysis of disease chromosomes can help identify probable historical recombination events and localize disease mutations. Most available analyses use only marginal and pairwise allele frequency information. We have developed a Bayesian framework that utilizes full haplotype information to overcome various complications such as multiple founders, unphased chromosomes, data contamination, and incomplete marker data. A stochastic model is used to describe the dependence structure among several variables characterizing the observed haplotypes, for example, the ancestral haplotypes and their ages, mutation rate, recombination events, and the location of the disease mutation. An efficient Markov chain Monte Carlo algorithm was developed for computing the estimates of the quantities of interest. The method is shown to perform well in both real data sets (cystic fibrosis data and Friedreich ataxia data) and simulated data sets. The program that implements the proposed method, BLADE, as well as the two real datasets, can be obtained from http://www.fas.harvard.edu/ ∼junliu/TechRept/01folder/diseq_prog.tar.gz.In the quest to identify genes responsible for specific illnesses, it has been observed in many cases that a large portion of the carriers of the disease gene in the current population are descendant from a small number of "founders" in whose genomes the deleterious mutation appeared some generations ago. This translates into inhomogeneity between the allele frequencies in the general population and those with the disease for genetic markers close to the location of the disease gene(s). The reason is that the allele frequencies of these markers in the disease population still reflect those originally carried by the founder chromosome(s), with modifications introduced by recombinations and mutations. This phenomenon, known as linkage disequilibrium (LD), can be exploited to identify the location of a disease gene by measuring the dependence between disease status and allele distributions among a set of markers.Simply looking at the marginal dependency between each marker and disease status in a case/control sample of chromosomes is clearly inefficient. For an LD mapping strategy to be optimal in fine mapping, it is essential to consider the information observed in a set of contiguous markers (i.e., haplotypes). The primary goal of our Bayesian analysis is the localization of a gene responsible for the disease within the considered set of markers. Secondary goals are the determination of ancestral haplotypes, the separation of distinct founders of the disease, the construction of haplotypes from unphased chromosomes, and inference on the ages of the mutations causing the disease. Our method, like any others based on LD, is appropriate when there are reasons to assume the existence of a founder effect in at least a significant proportion of the diseased individuals. We note that several attempts along the lines of our approach have been discussed in the literature, and we compare these methods with our approa...

Population history and infrequent mutations: how old is a rare mutation? GUCY2D as a worked example

“…Considering the entire haplotype leads to more robust estimates than pairwise disequilibrium. 19 Bayesian procedures require specification of prior distributions for all the parameters. 20 We implicitly assume in our analysis that the prior age is drawn from a uniform distribution.…”

Exhaustive allelic transmission disequilibrium tests as a new approach to genome-wide association studies

“…This finding is not unexpected because the most informative alleles are not necessarily the ones closest to the disease-causing mutation 20 . Other methodologies exist specifically to identify mutation location in the context of fine mapping [21][22][23][24] . Often, these algorithms assume that there is a single disease-causing mutation at a given locus or are unsuitable for genome-wide analysis because of their computational burden.…”

Experiences with GeneRecon on MiG

“…To approach this problem, we have developed an MCMC algorithm 1 based on Morris et al's shattered coalescent method [8], but incorporating ideas from Liu et al [4] and Molitor et al [7], where we separated the affected individuals into mutation clusters -where affected individuals in the same cluster are assumed to be descendants of a common founder -and a null cluster -for individuals affected due to environmental factors and not genetic factors. By sampling, during the run of the MCMC, the distribution of affected individuals among the mutation and null clusters, we hope to be able to infer which individuals carry an increased risk of mutation and which are affected solely due to environmental factors, while at the same time locating the locus of the disease affecting gene.…”