Abstract: INTRODUCTION & OBJECTIVES: The genomic heterogeneity between prostate cancer specimens has hampered clear correlation between the mutational landscape and the aggression or histopathological grade of the tumour. By gaining a consensus of the sequence of genomic events that manifest during tumourigenesis within individual tumours, important driver mutations may be found that both highlight additional therapeutic targets and increase prognostic accuracy. This paper presents the results from determining the temporal evolution of large scale mutations in localised prostate cancer through the collated analysis of large scale genomic databases. These results are collated by Gleason grade in order to reveal genes that are significantly and progressively altered. MATERIAL & METHODS: SNP array 6.0 and clinical data were taken from The Cancer Genome Atlas (TCGA) Research Network: http://cancergenome.nih.gov/. Copy number profiles were corrected for tumour purity and overall ploidy. The accuracy of mutations called within sub-clonal tumour fractions was enhanced by phasing SNPs together in parent specific haplotype blocks using data from the 1000 Genomes Project. Within each tumour sample sub-clonal copy number events were assigned to clusters that were defined using a hierarchical Bayesian Dirichlet process. The clonal fraction of each cluster determined their temporal order within each tumour. Copy number aberrations that were significantly enriched by permutation analysis were ordered in relation to other significantly enriched regions within each tumour (FDR, p {$<$} 0.05). This analysis was repeated over all tumour samples to form an overall order. This relied upon the Bradley-Terry model of pairwise comparison with bias reduced maximum likelihood. A list of somatic alterations implicated in the development of prostate cancer was compiled from relevant sources in the literature. Genes located within significantly enhanced regions from this analysis were ordered by their genomic loci. A modified χ2 test (p {$<$} 0.05) was applied to determine the relationship of significantly enhanced genes with Gleason score for both the fully clonal and sub-clonal copy number aberrations. RESULTS: Many of the large scale mutations cannot be separated temporarily, reflecting the heterogeneity seen in the mutational landscape of prostate cancer. However, there are a few notable exceptions: • Loss of heterozygosity occurs significantly prior to copy number gains, which in general occur significantly prior to homozygous deletions. • Homozygous PTEN loss is an early event in prostate cancer oncogenesis. • Deletion at the TMPRSS2-ERG locus occurs early in prostate cancer oncogenesis. Gleason score significantly affects both the fully clonal and sub-clonal mutational landscape with a number of tumour suppressor genes and oncogenes progressively enriched with tumour grade. These include genes closely involved in cell cycle, PI3K-Akt, RAS/ RAF, and androgen receptor signalling. CONCLUSIONS: The statistical analysis of large scale genomic datasets has allowed a unique analysis into the oncogensis of copy number aberrations in localised prostate cancer. This has revealed a consensus in the both the temporal order and the histopathological correlation of large scale mutations. The data implies that tumour grade evolves temporarily and is strongly associated with large scale mutational events.