View Program - 2022 Population, Evolutionary, and Quantitative Genetics Conference

Modern quantitative trait locus (QTL) mapping in panels of inbred strains uses a linear mixed model (LMM) to test for SNP-phenotype association while accounting for a random effect of population structure. A decade of mathematical tricks have mitigated the computational expense of repeatedly fitting this complex model for genome-wide applications. Existing procedures, however, make the assumption that the phenotype of each strain (or individual) is known with equal precision. In reality, this assumption does not always hold. We propose a method, weighted Inbred Strain Association Mapping (wISAM), which accounts for heteroscedastic residual variance in the study population using a weighted regression technique and makes use of variance shrinkage methods to stably estimate these weights. Simulation studies comparing wISAM to existing methods demonstrate that it can provide additional statistical power for GWAS. The method is then illustrated using data from studies incorporating the Hybrid Mouse Diversity Panel (HMDP).

Increasing power in inbred strain association mapping by recognizing variance heterogeneity