Computational Strategies for Deciphering the Genetic Architecture of Lung Cancer in Individuals of African Ancestry

Abstract

Lung cancer is the leading cause of cancer-related deaths in the United States. Despite smoking fewer cigarettes than White Americans, Black/African American individuals are disproportionately affected by lung cancer. In addition to social and environmental influences, genetic factors are thought to contribute to this disparity. In the first aim of this work, I conducted a comprehensive study of the genetic architecture of lung cancer in a cohort of 6,490 individuals of African ancestry, utilizing existing computational approaches. Next, to improve biological interpretation and downstream prioritization of genetic variant associations identified in Aim 1, I developed a decision tree-based computational framework capable of identifying functionally relevant genetic variation in disease-associated cell and tissue types and linking these predicted regulatory variants to their putative target gene(s). Finally, in the third aim, I trained elastic net regression-based models of genetically regulated enhancer RNA (eRNA) expression across 49 human cell and tissue types. I demonstrate the utility of these models in identifying understudied contributors to the genetic architecture of lung cancer in individuals of African ancestry. Collectively, these results reveal unique genetic contributors to lung cancer risk in individuals of African ancestry and provide a novel set of computational resources for improving future genetic research in African ancestry populations.