Analysis of Extended Pedigrees Ascertained for Asthma

This dissertation describes statistical analyses of an extended pedigree resource ascertained for asthma. The asthma pedigree resource contains 741 affected individuals in 81 pedigrees, with approximately 90 percent of the resource genotyped on 535 short tandem repeat markers dispersed throughout the genome. As a motivation for our analysis of the asthma pedigree resource, we first establish strong evidence for a heritable component to asthma using extensive population genealogical and phenotypic data from a combined resource of the Utah Population Database and the Utah death certificate registry. Subsequent analysis of the asthma pedigree resource itself consisted of various strategies designed to map the genomic locations of predisposition loci for asthma. First, we conducted a genomewide scan using a parametric linkage statistic and two general models. We interpreted the results of the scan using thresholds based on the asymptotic distribution of a related statistic and concluded that chromosomal region 5ql3 is linked to asthma under a recessive mode of inheritance. Second, recognizing that the selected thresholds used to interpret the outcomes of the first scan may be unreliable and that an empirical evaluation could be informative, we applied the latent p-value method for linkage to a genomewide scan of the resource. The latent p-value method offers fast empirical assessment, compared to conventional methods, and the flexibility to consider any linkage statistic without reference to any theoretical or asymptotic distribution. The latent p-value analysis confirmed the presence of linkage to chromosome 5ql3, albeit with a different linkage statistic than was used the in the original genomewide scan. Third, recognizing that consideration of only two general models may be insufficient to adequately model the true mode of inheritance at a linked locus, we conducted another genomewide scan using the modscore statistic, with the benefit of empirical assessment from the latent p-value method. The mod score analysis exhibited very low power to detect linkage and did not identify any new loci of interest. Taken together, our analyses indicate that there is linkage in this resource to chromosomal region 5ql3 and further expansion of this resource is warranted.