A novel approach to gene-set enrichment for two-class time-course expression data

In this paper we describe a novel approach to gene-set enrichment, specifically designed for two-class time-course expression data. We first define a space of differential gene expression (DEG), derived from the original gene data space. Every gene-set is then scored, computing the sum of the dot-products among its genes, encoded as vectors in the DEG space. The score is tested for significance versus a random dot-product sum distribution, generated using randomly sampled gene sets. We display the enrichment results obtained for a two-class time-course microarray data-set, profiling the transcriptomes of wild type mice versus transgenic mice undergoing dilated cardiomyopathy; the enrichment results are highly coherent with a-priori knowledge on the pathogenetic process. As a preliminary assessment of the comparative performance, we show that the dot-product method outperforms the pre-ranked mode of the well-established enrichment method GSEA.