Compared to gene set enrichment analysis, the SNP set enrichment analysis is a much larger scale and is more computationally challenging. Several pathway-based methods have recently been developed to analyse GWAS [13,20,41-44]. In general, these methods can be classified into two categories, depending upon how representative SNPs for each gene are chosen: one selects the most significant SNP per gene, and the other selects all SNPs located within a gene [20]. Both approaches have limitations. Using all available SNPs per gene not only poses computational challenges, but also introduces significant amounts of noise into the analysis. Using the most significant SNP per gene might miss SNPs with moderate strength individually but strong effects jointly, and in addition it introduces biases of favouring large extensive pathways and genes with greater numbers of SNPs. The SSEA method we proposed uses an adaptive approach to choose SNPs in each gene, and can eliminate the limitations of other strategies.
