Investigating the microbial basis of early childhood caries via integrative analysis of metagenomics metatranscriptomics and metabolomics

The increasing availability and scale of omics data have revolutionized our ability to understand complex biological processes underlying health and disease. Such biologically informed insights are aligned with the notion of precision medicine and have the potential to improve diagnoses, prevention, and treatment. In the oral health domain,

multiple omics data layers (e.g., genomics, metagenomics, transcriptomics, metabolomics), intended to capture aspects of otherwise unobservable biology, are increasingly being collected in oral health studies. However, methods for powerful and informative integration of information gained from these multiple data layers remains

elusive. The focus of this proposal, early childhood caries (ECC), is the most common non-communicable childhood disease. ECC is defined as dental decay in children under the age of 6. It remains a clinical and dental public health problem and confers substantial and multi-level human and economic impacts. The advent of precision oral

health care, based upon a new, microbially-informed understanding of ECC, is expected to shed light onto mechanistic aspects of the disease processes and reveal new ways to prevent it. We propose to develop new statistical methods and machine learning (ML) strategies that handle high dimensionality and excess zeros in comprehensive

integrative analyses of metagenomics (MTG), metatranscriptomics (MTX), and metabolomics (MTB) and apply them to characterize the ECC-related supragingival biofilm dysbioses, biochemical activities, and their interactions in a community-based sample of 300 preschool-age children, enrolled in a large-scale investigation of early

childhood oral health in North Carolina. We propose to develop statistical methods to jointly identify regulatory modules across MTG, MTX, and MTB associated with ECC. Furthermore, we seek to develop interpretable machine learning approaches to predict binary (e.g., case status) or quantitative ECC experience (e.g., dmfs index) using MTG,

MTX, and/or MTB. We anticipate that our study will provide novel insights into the microbial basis of ECC, identify microbial and biochemical disease biomarkers, highlight potential targets for prevention and treatment, and establish a powerful research platform that can be extended to characterize other microbiome-related diseases.