Skip to main content
. 2022 Jun 9;7:33. doi: 10.1038/s41525-022-00303-2

Fig. 3. Manhattan plots for five genome-wide association studies (GWAS) of drug-induced change in low-density lipoprotein cholesterol (LDL-C) from IMPROVE-IT PGx study.

Fig. 3

These five Manhattan plots are from Model 1 (a, M1), Model 5 (b, M5), Model 3 (c, M3), Model 6 (d, M6) and Model 7 (e, M7). M1 used log-fold-CFB as phenotype, adjusted for baseline LDL-C and used a 2-step approach in the regression, in which residuals were obtained by regressing out the covariates and then inverse normally transformed. M3 was the same as M1 except that it did not adjust for baseline LDL-C in the model. Both M1 and M3 yielded the same three significant loci. M5 used log-fold-CFB as the phenotype, adjusted baseline LDL-C, and used the 2-df test in the regression. M6 was the same as M5 except that it did not adjust for baseline LDL-C in the model. M5 yielded three significant loci while M6 yielded two additional significant loci (STAG1/SLC35G2/NCK1, SLCO1B1) on chromosome 3 and 12, respectively. M7 used log-baseline as the phenotype for the baseline association test, which yielded one significant locus on chromosome 19. The horizontal red line represents the whole-genome significant p value threshold 5e−08. All tests were two-sided.