A GS-CORE algorithm for performing a reduction test on multiple gene sets and their core genes

Gene-set analysis seeks to identify enriched gene sets that are strongly associated with the phenotype. In many applications, only a small subset of core genes in each enriched gene set is likely associated with the phenotype. The reduction of enriched gene sets to the corresponding leading-edge subsets of core genes is a useful way for biologists to understand the biological processes underlying the association of a gene set with the phenotype of interest. Therefore, we propose a new gene-set analysis that tests the significance of enrichment on multiple gene sets, while simultaneously determining the corresponding leading-edge subsets of core genes. In the proposed analysis, we assigned a newly defined enrichment score to each gene set, and then corrected the statistical significance of the score for multiple testing of many gene sets by controlling the false-discovery rate.