A multiple regression method for genomewide association studies using only linkage information

Abstract

Genomewide association studies (GWASs) typically require a base of linkage disequilibrium (LD) to capture quantitative trait locus (QTL) signals. In this study, we tested whether identifying QTLs in the framework of GWAS can be based only on linkage information. Our study sought to validate a method to replace LD with linkage in association studies, and we investigated the statistical power of different heritabilities and the number of QTLs using simulation data. We found that it is entirely feasible to exploit the multiple regression method for GWASs using only linkage information. Similar to the typical genomewide association tests using LD information, our new approach performed validly when themultiple regression based on linkage method was employed. However, the performance improved slightly when the linkage was used alone, which was much closer to the traditional GWAS model using single marker regression. Meanwhile, the statistical power of the new method decreased with increasing number of QTLs, and its power was sensitive to heritability. In summary, these results suggest that this method can identify QTLs, although the power is relatively weak. The cause of this phenomenon remains unknown.