Binary classification machine-learning Unveils Sex-Dependent mutated gene Signatures in Melanoma

Abstract

There are significant differences in the prevalence of cancer type, primary tumor body site, and mutation load between men and women, but the mechanisms underlying these sex-dependent differences is mostly unknown. Here we used binary classification machine-learning methodology to study sex-correlated somatic mutations signatures in cutaneous melanoma. We identified a number of genes that are more frequently mutated in females compared to males. Mutations in two genes, LAMA2 and TPTE, together with a set of specific genes that are not mutated, can predict sex of melanoma patients. Over representation analysis of genes clustered with LAMA2 revealed significant enrichment in androgen and estrogen biosynthesis and metabolism pathways, suggesting that mutation of LAMA2 might be involved in biased sex hormone synthesis in melanoma. Taken together, our analysis shows that gender can be predicted based on mutation status of genes in melanoma and that certain mutations are predictive of survival beyond sex differences. Our results will lead to better diagnosis and more effective personalized treatment of melanoma. SummaryIt was observed that between men and women there is a significant difference in the prevalence of cancer type, the primary tumor body site, and the number of mutations found in a given tumor type. However, the mechanisms behind these gender differences are mostly unknown. To investigate sex corelated somatic mutation signatures in cutaneous melanoma we used binary classification machine-learning methodology. We identified specific genes that are more frequently mutated in females compared to males, including LAMA2 and TPTE, which are predictive of gender. We also found a significant enrichment in androgen and estrogen biosynthesis and metabolism pathways clustered with LAMA2, suggesting that mutation of LAMA2 might be involved in biased sex hormone synthesis in melanoma. We showed that gender can be predicted based on mutation status of genes in melanoma and that certain mutations are predictive of survival. Our findings could lead to better diagnosis and more effective personalized treatment of melanoma.