Metascape utilizes the well-adopted hypergeometric test58 and Benjamini-Hochberg p-value correction algorithm59 (link) to identify all ontology terms that contain a statistically greater number of genes in common with an input list than expected by chance. By default, Metascape pathway enrichment analysis makes use of Gene Ontology16 (link), KEGG17 (link), Reactome18 (link), MSigDB19 (link), etc. Distinguishing it from many existing portals, Metascape automatically clusters enriched terms into non-redundant groups, where it implements similar logic as found in DAVID6 (link). Briefly, pairwise similarities between any two enriched terms are computed based on a Kappa-test score28 (link). The similarity matrix is then hierarchically clustered and a 0.3 similarity threshold is applied to trim the resultant tree into separate clusters. Metascape chooses the most significant (lowest p-value) term within each cluster (Supplementary Data 4 ) to represent the cluster in bar graph and heatmap representations. The analysis provides other popular enrichment metrics in addition to p-values.
Free full text: Click here
