Fig. 1
From: MARS: leveraging allelic heterogeneity to increase power of association testing
Overview of MARS. Here, we assume that we are testing an association between a locus of m variants and a trait. The leftmost panel shows the input of MARS; m number of summary statistics for the locus and an n×m matrix that contains genotypes of m SNPs for n samples. The next two panels on the bottom show the re-sampling process in which we sample the null statistics K times from an MVN distribution with a variance-covariance matrix of Σ that contains LD of the genotypes X. The rightmost panel shows the process by which we estimate LRTstats for the null panel from which we can compute a p-value for the data