In cancers with a highly altered genome, distinct genetic alterations drive subsets rather than the majority of individual tumours. Here we use a sequential search across human tumour samples for transcript outlier data points with associated gene copy number variations that correlate with patient's survival to identify genes with pro-invasive functionality. Employing loss and gain of function approaches in vitro and in vivo, we show that one such gene, MTSS1, promotes the ability of

more »

... cells to metastasize and engages actin dynamics via Rho-GTPases and cofilin in this process. Indeed, high MTSS1 expression defines a subgroup of primary melanomas with unfavourable prognosis. These data underscore the biological, clinical and potential therapeutic implications of molecular subsets within genetically complex cancers.