the causative gene for only the aromatic hydrocarbon response could be identified using EMMA [14], because it is a binary response (either present or absent) phenotype. If a binary phenotype is entirely determined by alleles at a single SNP, then GWAS and HBCGM will have the same detection power. For example, conventional GWAS methods have been used to identify causative genetic factors for some traits in outbred strains of mice [15]. However, the poor performance of EMMA when analyzing the H2-Eα gene expression and survival after fungal infection data was striking: EMMA identified >516,000 SNPs (corresponding to 783 Mb of the genome) with a higher correlation than the causative gene effecting survival after fungal infection, and ~10,000 SNPs had a higher correlation with the H2-Eα gene expression data in Figure 2 than the experimentally verified cis-acting SNPs within H2-Eα. Since EMMA can analyze only one SNP at a time, it is not the optimal method for analyzing phenotypic traits with 3 discrete phenotypic states. Although there are many situations were HBCGM will be unable to identify the genetic basis for trait differences, it can be productively utilized to evaluate traits that are controlled by multiple polymorphisms within a contiguous region,
