A haplotype-linkage analysis method for estimating recombination rates using dense SNP trio data

Meiotic recombination is an important evolution force in shaping human genomes and creating human diversity. Recent explorations suggest that meiotic recombination events tend to happen in certain regions of the genome, leading to the hypothesis of recombination hot spots. To identify such hot spots, we describe an empirical method for estimating the recombination rate, which reflects both historical and current meiotic events, using unphased genotypes from nuclear families, in particular, parents-child trios. The key idea is that use of the haplotypic polymorphisms with multiple adjacent SNPs will increase the number of informative meioses and hence would improve the power of linkage analysis. Since haplotypes of individuals are not directly observed, e.g., in HapMap data, we infer the haplotypes simultaneously while estimating recombination rate. We refer this described method as haplotype-linkage (HALIN) method. Our simulation results show that HALIN gives unbiased estimates of recombination rate. We apply HALIN to analyze the genotype data of chromosome 20 from HapMap data and compare the results to the results using two existing methods, Bayesian coalescent method described by McVean et al. [2004] science 304:581-584 and empirical method by Clarke and Cardon ([2005] Genetics). Results suggest that HALIN identified 75 hot spots on chromosome 20, and 85% of them are also identified by the method by McVean et al. ([2004] science 304:581-584), in addition to a few new hot spots. In comparison with Clarke and Cardon's result, estimated recombination rates (RRs) under these 75 hot spots are significantly greater than those outside of hot spots, which support the general consistency in discovering recombination hot spots between two methods.