Annotation data

We collect pathway annotation data assigning genes to pathways, from the Reactome pathway ontology [18]. For each of our five molecular networks, we create a set of pathway networks by inducing the gene set of each pathway on the network. For each molecular network, we consider those pathways that, once induced on the full network, form a connected subgraph of a size of at least 10 and up to 100 nodes. We provide the distribution of pathway sizes for each of our molecular networks in S1 Fig in S1 File. The number of pathways considered per molecular network is summarized in S2 Table in S1 File.

We collect experimentally validated functional annotations from the Gene Ontology (i.e., evidence codes ‘EXP’, ‘IDA’, ‘IPI’, ‘IMP’, ‘IGI’, ‘IEP’), that assign genes to biological process annotations (GO-BP), cellular component annotations (GO-CC) and molecular function annotations (GO-MF) [24].

We collect 586 cancer driver annotations from the intOGen database [25]. We consider a gene to be a cancer driver if it is a known cancer driver in at least one cancer type.