Certain common diseases and traits may be more amenable to GSEA approaches than others, depending on their genetic architecture. In addition to identifying new biological pathways or processes associated with disease risk or trait variation, GSEA methods, such as MAGENTA, may provide predictions for new disease or trait genes of modest effects (top ranked gene scores in enriched gene sets). Such joint analysis of SNPs (or other types of variants) at the gene and gene set levels should be most useful for detecting associations in a narrow range of nominal significance levels (between noise levels, e.g. p<0.1, and SNP replication cutoff, e.g. p>∼0.0001), a range that has been shown to contain associations of small effect in polygenic disorders [53]. The GSEA approach may also help prioritize potential causal genes in validated association regions that contain multiple genes.
