Development of a novel GWAS method to detect QTL effects interacting with the discrete and continuous population structure

Although GWAS has been a key technology to identify causal genes, the current standard GWAS model still has problems that need to be solved. Among them, the population structure is one of the most severe problems when detecting QTLs in GWAS since the GWAS model is statistically confounded by effects derived from the population structure. Further, the existence of QTLs, whose effects depend on the genetic background, also affects the conventional GWAS results by causing many false positives. Although the model to detect these population-specific QTLs has already been developed, this model requires prior information on the population structure, which may only sometimes be available. Also, the previous model only assumed the situation where QTLs interact with the discrete population structure. However, target populations of GWAS often consist of genetic resources with a more continuous population structure, and there has been no model that can consider such QTLs interacting with the continuous structure. In this study, by explicitly including an interaction term between a SNP/haplotype block and the genetic background in the conventional SNP-based/haplotype block-based GWAS model, we developed two models, named SNPxGB and HBxGB, that can detect QTLs interacting with the discrete and continuous structure. Our developed models were compared to the previous models by a simulation study assuming some types of QTLs, i.e., QTLs with effects common to all the backgrounds, specific to one genetic background, and interacting with polygenes. The simulation study showed that the models assuming the same situation as the simulation settings for each QTL type were suitable for detecting the corresponding QTLs. Primarily, our second HBxGB model could detect QTLs interacting with polygenes, i.e., continuous population structure, better than the previous model utilizing the prior population structure information. Our developed models are expected to help unravel the unknown genetic architecture of many complex traits. ### Competing Interest Statement The authors have declared no competing interest.