Fig. 1
From: Identifying latent genetic interactions in genome-wide association studies using multiple traits
Overview of the Latent Interaction Testing (LIT) framework. Given a set of r traits, \(\varvec{Y}\), and \(m_{0}\) SNPs, \(\varvec{X}\), the goal is to detect a latent genetic interaction involving the SNPs. The trait squared residuals (SQ) and cross products (CP), \(\varvec{Z}\), are calculated while adjusting for linear effects from the genotypes and any other covariates. The traits and genotypes are also adjusted for population structure. A similarity matrix for the genotypes, \(\varvec{K}_{X^{\prime }}\), and the SQ and CP, \(\varvec{K}_{Z}\), are calculated to construct a test statistic, T, which measures the overlap between the two matrices. Large values of T are evidence of a latent genetic interaction