Abstract:
|
Many gene expression studies look at whether a treatment has similar or opposing effects compared to other treatments (e.g. the effect of a gene deletion vs. overexpression on other genes). Often, researchers seek to understand the impact of these treatments on biological functions by functional enrichment analysis (testing if disrupted genes are overrepresented in a biological pathway). Statistical methods specifically for equivalently or opposingly changed genes are lacking, so researchers often resort to simple comparisons of directionality of change across treatments/experiments. In this work, I introduce the Equivalent Change Index (ECI), which measures the degree of equivalent or inverse change across treatments or experiments, and propose statistical tests for equivalent change at the level of genes, pathways, and genome. Using simulated and biological data, I show the ECI and related tests are an effective means of detecting equivalent and inverse change in gene expression and demonstrate a novel method of analysis that can detect pathways enriched for genes in cases when typical GSEA cannot, allowing a focus on changes that are particularly relevant to certain questions.
|