Figure 1
From: New insights into old methods for identifying causal rare variants
Selection by sliced inverse regression (30 markers). For group 1, using 30 top markers, the absolute values of the first SIR direction coefficients are plotted in sequence. Dashed lines indicate the answer SNPs. Two-mean clustering is performed on the absolute coefficients. The red circles represent the markers in the cluster with a higher mean, which are our final selections. The green circles represent markers in the other cluster. Other groups produce similar plots. In this case, five out of five selected markers are answers.