Overview of the HotNet algorithm. HotNet uses a heat-diffusion process to identify significantly mutated subnetworks within an interaction network. (a) Heat is assigned to each gene according to the proportion of samples containing a single-nucleotide variant (SNV) or copy number aberration (CNA) in the gene. (b) The initial heat then spreads on the edges of the network for a fixed amount of time. Removing cold edges connecting genes that do not exchange large amounts of heat breaks the network into smaller subnetworks. (c) HotNet assesses the number and size of the resulting subnetworks using a two-stage statistical test.