Bioinformatics Seminar
Time: 11AM
Venue: Teams
21 June 2022
A nutrition aware model to identify diet-microbiome-host health interactions
Jean YangUniversity of Sydney
The human body is home to complex microbial communities. In recent years, considerable evidence has emerged indicating that the microbiome is an important contributor to an individual's health. Furthermore, the gut microbiome is known to change throughout our lives as a result of various environmental influences. Diet is one of these factors, and different diets can affect relationships between microbiome composition and host health/disease status. Here, we present the Nutrition-Ecotype Mixture of Experts (NEMoE) model that accounts for uncovering associations between the gut microbiome profile and the health state of an individual that considers diet-specific cohort heterogeneity. We evaluate the performance of NEMoE in both simulation and real-world data. The results show that NEMoE is capable of not only improving prognostic performance for detection of disease but also for identifying diet-specific microbiome markers. Next, we will explore the potential of incorporating existing strategies that can predict sample outcomes in a reference-free manner, thus addressing the implementation barrier of microbiome markers in practice.