Power and robustness of model-free multipoint linkage analysis

This study compared the power and robustness of the multipoint preliminary linkage analysis approach for quantitative traits propose by Goldgar to the nonparametric Haseman-Elston sib-pair test and a parametric multipoint analysis package, LINKAGE. Based on the proportion of genetic material shared identical be descent (IBD) by sibling, this multipoint method, MIBD, detects regions of chromosomes contribution to quantitative trait variance. Monte Carlo simulation methods were used to generate genotype, phenotype and marker information for nuclear families over specified chromosomal models. The influence of differing degrees of trait heritability, gene frequency, and trait variability due to genes at loci in the test region was investigated. The effects of multiple trait loci or polygenes, trait locus dominance, and shared sibling environment were also investigated. Additional topic evaluated in this study included the absence of parent phenotype measures and the effects of marker information. The results showed that the MIBD method is consistently more powerful than the Haseman-Elston sib-pair test in detecting linkage over a wide rage of inheritance models and trait parameter variation while retaining equal levels of significance. In comparison the parametric LINKAGE analysis, the MIBD method yield lower results for major trait locus models when parental phenotypes were known, but showed higher power for detecting linkage when parental phenotypes were unknown. MIBD analysis also yielded better results than LINAHE for polygenic trait models and unlinked trait models. MIBD analysis ran much faster than LINKAGE analysis further demonstrating the feasibility of using the MIBD method as a preliminary search tool. In conclusion, the MIBD method proves a powerful, robust means of analyzing linkage with minimal restrictions on model assumptions.