In this supplementary, we demonstrate the heatmaps of semantic similarity matrices from different semantic measures, which are “Rel”, “Resnik”, “Lin”, “Jiang” and “Wang” (see the GOSemSim vignette for more details of the methods). The matrices are calculated based on 500 randomly sampled GO terms, from Biological Process (BP), Molecular Function (MF) and Cellular Component (CC) ontologies. For each ontology category, we show 5 examples.
Generally, the similarity matrices generated by the “Rel”, “Resnik” and “Lin” methods show more clear diagonal block patterns.