Fig. 1
From: Detecting gene–gene interactions from GWAS using diffusion kernel principal components
Novel epistasis detection workflow: SNPs are mapped to respective genes followed by imputation for missing genotypes. Interaction information between SNPs allocated to the same gene is used to compute diffusion kernels and graphical within-gene network structures. Data reduction via kernel principal component analysis gives gene summaries (representations) that are submitted to an epistasis detection model of choice (here, for illustration, a Bayesian model). Gene-gene interactions are graphically represented via a gene-level statistical epistasis network and interpreted