"Tumorigenesis is a form of somatic evolution, a topical subject given the advent of cancer genome sequencing. However, we contend that some features of Darwinian evolution have been neglected when cancer is studied, as have some aspects of evolution that are special to cancers. For example, tumours comprise an expanding population of cells, cancers must occur within a normal human lifespan, and genotypes detrimental to growth of the tumour as a whole may be selected. These factors may render invalid the classical model in which successive mutations with large advantages arise and spread through the tumour in selective sweeps. To incorporate these neglected features and to test how tumorigenesis depends on factors such as mutation rate, selection and size constraints, we shall set up a comprehensive model of tumour growth incorporating cell birth, death, division and mutation parameters. We shall examine specific aspects of cancer-as-evolution in mice. By marking mutant clones using fluorescent proteins, we can track them and see how they persist, spread and die. We shall also determine the mutation profiles and genetic diversity of mutant clones and whole tumours in mice and humans using next-generation sequencing. Specific experiments will determine: (i) the fate of new advantageous clones arising in an existing tumour; (ii) whether new disadvantageous clones can persist in tumours; (iii) whether apparently maladaptive traits for tumour growth, such as suppressing the growth of competitors, can be selected; (iv) why do housekeeper gene mutations cause cancer in specific sites; (v) can cancer cells have too much genomic instability; and (vi) whether all cancers develop owing to driver mutations with big effects, or are there “mini-drivers” of tumorigenesis? There will be continual cross-talk between the experimental and modelling work. The results of the project will enhance our basic understanding of tumorigenesis and suggest strategies for anticancer therapy."