Abstract Cutaneous melanoma is an aggressive form of skin cancer. Half of patients with advanced localized melanoma will die (stages IIA-IIIB) form the disease. The prognosis among these patients is highly variable. Broad risk classification based on clinical factors has limited predictive power at the level of individual patients and has no direct therapeutic implications. There is a great need to identify novel biomarkers for more refined classification of patients into different risk categories to allow for enhanced surveillance and to guide the use of adjuvant therapies for early intervention. The program project aims to assemble a large cohort of melanoma primary tumor samples from over 1,000 patients across nine different institutions with long-term follow-up and detailed clinical and pathological information. Genomic, epigenomic, and transcriptomic profiling of these tumors and in-depth analyses of the individual platform will be carried out in Project 1-3 respectively. The goal of this Project (Project 4) is to conduct systematic analyses across these data types for multiplatform classification of melanoma subtypes and prognosis. In Aim 1, we will identify integrated molecular subtypes in primary melanoma jointly defined by somatic mutation, DNA copy number, DNA methylation, mRNA and miRNA expression. The subtypes will be correlated with clinical, pathological variables and survival outcome. In Aim 2, we will synthesize features across different platforms to derive an integrated prognostic signatures from the discovery cohort (n=660), benchmark the performance against clinical variables, and validate the prognostic value of the integrated signature in the test set (n=340). In Aim 3, we will conduct an intra-tumor heterogeneity analysis to reveal the clonal status of melanoma driver genes and the spectrum of clonal heterogeneity among a large cohort of primary melanoma as potential biomarkers for prognosis and to inform therapeutic strategies.