a) Signature 3 tumors had significantly elevated numbers of LoH (p = 0.005, Kolmogorov-Smirnov, two-sided; p = 5.34 x 10−5, univariate logistic regression, two-sided), TAI (p = 4.4 x 10−5, Mann-Whitney U, two-sided), and LST (p = 0.007, Mann-Whitney U, two-sided) events. b) Signature 3 was also enriched in TWT tumors of two independent melanoma WGS cohorts, suggesting that the assignment of signature 3 was not ambiguous or the result of noise from lower mutational load in WES data. Asterisks denotes a Mann-Whitney U p-value < 0.05. c) Although indel signature ID8 was found in the majority of melanoma WGS samples, the relative contribution of indel signature ID8 was significantly higher in TWT tumors (p = 3.3 x 10−3, Kolmogorov-Smirnov, two-sided). d) Significance vs. effect size (fold-change) of significantly differentially expressed (Benjamini-Hochberg, q-value cutoff < 0.05; signature 3 vs. non-signature 3) DNA repair genes via edgeR. Gene names highlighted in purple function in DSB repair pathways including HR. e) The distribution of expression between putatively DSB repair deficient and non-DSB repair deficient TWT tumors for DSB repair genes that were significantly differentially expressed (Mann-Whitney U; ordered by two-sided p-value, increasing, left-to-right, asterisks denote p < 7.6 x 10−3). The data is represented as a boxplot where the middle line is the median, the lower and upper edges of the box are the first and third quartiles, the whiskers represent the interquartile range (IQR) multiplied by 1.5, and beyond the whiskers are outlier points.