Fig. 7From: Identifying novel associations in GWAS by hierarchical Bayesian latent variable detection of differentially misclassified phenotypesAssociation analysis of epilepsy dataset using PheLEx. Manhattan plots (x-axis: SNP genomic position, y-axis: -log10p-values of association test > 2) of (a) GWAS results for original epilepsy phenotype with Bonferroni-corrected p-value threshold shown as dark gray line, (b) GWAS results for PheLEx corrected epilepsy phenotype (where PheLEx-identified cases are switched to controls). Training SNPs used as input for PheLEx are marked in blue whereas differentially significant SNPs are marked in red. Differentially significant SNPs are defined as SNPs not included in training PheLEx that are statistically significant using Benjamini-Hochberg procedure (adjusted p-value < 0.1) in association analysis with the PheLEx corrected epilepsy phenotype and not with original epilepsy phenotype. Manhattan plots show linkage disequilibrium (LD) pruned SNPs onlyBack to article page