CFE 2019: Start Registration
View Submission - CMStatistics
Title: Adapting methods of gene set enrichment analysis to study of the human microbiome Authors:  Levi Waldron - CUNY Graduate School of Public Health and Health Policy (United States) [presenting]
Abstract: Despite advances in methods for differential abundance analysis in microbiome studies, biological interpretation of such results remains challenging. In a recent study of the oral microbiome in oral rinse specimens of heavy smokers, never smokers, and individuals likely exposed to second-hand smoke from the New York City Health and Nutrition Examination Study, we observed strong evidence of community-wide microbiome shifts correlated with primary and second-hand tobacco exposure. However, mechanistic interpretation of these shifts was challenging. To test the hypothesis that smoking affects oral microbiota based on their oxygen requirements, we annotated all bacterial genera observed in our study as aerobic, anaerobic, or facultatively anaerobic, creating corresponding microb sets for microbe set enrichment analysis. This analysis identified depletion of aerobic genera in smokers to be highly significant (p < 0.001, permutation test) which, combined with more traditional methods of causal inference, strongly suggest differentially harmful effects of cigarette smoke on the aerobic oral microbiome, rather than effects of unmeasured confounding. Our current efforts are discussed aiming at curating a broad base of microbial signatures to enable the development of methods for systematic microbe set enrichment analysis. New challenges and areas of research for statistical methods development are presented.