Pancreatic cancer (PC) is the fourth leading cause of cancer death with an overall 5-year survival rate of < 5%, a statistic that has changed little in almost 50 years. A deeper understanding of the underlying molecular pathophysiology is expected to advance the urgent need to develop novel therapeutic and early detection strategies for this disease. Genomic characterisation of PC has previously relied on targeted PCR based exome sequencing of small cohorts of mixed primary and metastatic lesions propagated as xenografts or cell lines (Jones et al, Science 321:1801-1806), leaving the true mutational spectrum of the clinical disease largely unresolved. Here we use exome sequencing (https://www.ebi.ac.uk/ega/studies/EGAS00001000154) and copy number analysis (not submitted) to define genomic aberrations in a prospectively accrued clinical cohort (n = 142) of early (Stage I and II) pancreatic adenocarcinoma. Detailed analysis of 99 informative tumours identified 1982 non-silent mutations and 1628 significant CNV events, and defined 439 significantly mutated genes based on stringent Significant Mutated Gene or GISTIC analysis. Integration with functional data from in vitro shRNA and in vivo Sleeping Beauty-mediated somatic mutagenesis screens provided supportive evidence for 184 of these as candidate driver mutations. Pathway based analysis recapitulated clustering of mutations in core signalling pathways in PC, and identified multiple new components in each, particularly in DNA damage repair mechanisms (ATM, TOP2A, TLM, RPA1). We also identified frequent somatic aberrations in genes involved in novel mechanisms including chromatin modification (SWI/SNF complex members, SETD2, EPC1), and axon guidance (Semaphorin, Slit, Netrin and Ephrin signalling), extending the number of core perturbed pathways in PC. Aberrant expression of axon guidance genes co- segregated with poor patient survival, and in animal models was associated with disease development and progression, further implicating perturbation of the axon guidance pathway as a novel mechanism important in PC. This dataset includes gene expression data from 90 primary tumour samples, 88 of which were used in this manuscript for survival analysis. Much of this data is also available through the International Cancer Genome Consortium (ICGC) Data Portal (http://dcc/icgc.org), under the project code: "Pancreatic Cancer (QCMG, AU)". Access to the strictly restricted clinical data must be made through the ICGC Data Access Compliance Office (http://www.icgc.org/daco). This dataset contains expression array data from 90 primary pancreatic ductal adenocarcinoma samples. One sample is present with two biological replicates, all others have 1 biological replicate.