Abstract Introduction: While targeted- and immuno-therapies have revolutionized the treatment of melanoma, a significant proportion of patients develop resistance to therapy and lack targetable tumor drivers. Thus, there remains a critical need for the discovery of novel tumor drivers to identify additional therapeutic targets. Melanoma has been divided into 4 genomic subtypes based on the presence of mutations in the three most frequently mutated, mutually exclusive, driver genes: BRAF, NRAS, NF1, and triple wild-type (TWT), and, in particular, no significantly mutated genes (SMGs) have been identified in TWT tumors. We hypothesized that uniform genomic analysis of expanded cohorts would provide power to identify driver genes altered at low frequencies, identify novel biological targets and therapeutic vulnerabilities unique to genomic subclasses, and specifically enable discovery of additional molecular properties unique to TWT melanoma. Methods: We aggregated 1,013 melanoma tumor and matched germline samples that passed joint quality control metrics and performed harmonized genomic analysis. Mutational significance analysis was performed using MutSigCV2 and OncodriveFML, and candidate genes filtered using TCGA expression data. Mutational signature analysis was performed using deconstructSigs, and validated in a separate cohort of melanomas. DESeq2 and edgeR were used to perform differential expression analysis. Results: We identified a set of 38 high-confidence drivers in melanoma (MutSigCV2, q < 0.1; OncodriveFML, q < 0.1), 21 of which were in either the COSMIC Cancer Gene Census or OncoKB. Several of these genes (e.g. ARID1A, CDK4) have not been considered significantly mutated or previously reported in melanoma. Further, mutational significance analysis within each subtype identified 63 BRAF, 73 NRAS, 26 NF1 and 11 TWT SMGs (MutSigCV2, q < 0.1), and revealed that secondary driver genes are rarely shared between the subtypes. Mutational signature analysis determined that a subset of melanomas (7%) have homologous recombination (HR) deficiency (signature 3), and this signature is significantly enriched in TWT melanomas (21.6%, p = 1.21 x 10-10, Fisher’s). ATM, APLF and SMARCD1 expression were significantly correlated with signature 3 contribution (p < 0.05), and significantly differentially expressed between signature 3 and non-signature 3 tumors in TWT melanoma (q < 0.05). Conclusion: Through uniform analysis of the largest melanoma whole-exome-sequenced cohort to date, we identified a set of high confidence driver genes, many of which were not previously associated with melanoma, as well as secondary driver genes unique to each genomic subtype. Mutational signature analysis identified the enrichment of HR deficiency in TWT melanomas, which appears to be facilitated through downregulation of ATM and may have immediate translational implications for patient-stratification of therapies that target HR deficient melanomas. Citation Format: Jake Conway, Amaro Taylor-Weiner, Saud AlDubayan, Brendan Reardon, Felix Dietlein, David Liu, Eliezer M. Van Allen. The secondary genomic architecture of melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3376.