Functional Interactions and Global Diversity in Gut Microbiomes: Advancing Microbial Community Type Identification with Deep Learning Approaches

Access to EGA study: EGAS00001001704 / Lifelines-DEEP

Development of computational “deep learning” methods for identifying and characterizing different types of microbial communities. Our study has currently processed and analyzed ~10.5K publicly available adult gut microbiome profiles. If we are to learn a robust model of human gut microbial community types, we require a large and comprehensive collection of human gut metagenomic profiles. The raw Lifelines DEEP Data data will be processed to further expand this compendium and provide more power to better model gut microbial community types especially when using whole-metagenome profiles. We aim to integrate and harmonize profiles from other recent datasets including several recent microbiome studies of under-represented populations. Microbial community types will be further analyzed to investigate their associations to the host’s attributes (age, gender) and metabolic profiles.

My research from 2014 onwards focused on the development of computational and genomic approaches for mapping molecular-level interactions between complex tissues (i.e. host systemic immunity and tumor tissue). Some challenges associated with this work are also similar to my previous research where the goal was to identify and characterize patient subtypes of breast cancer. This project re-directs this expertise towards understanding the functional variability and structure in human gut microbial communities across diverse populations.