Humans have co-evolved with complex, dynamic microbial communities that play essential roles in nutrition, metabolism, immunity, and numerous other aspects of human physiology. Hence, maintenance and recovery of key beneficial services by the microbiota in the face of disturbance is fundamental to health. Yet, stability and resilience vary in, and between individuals, and are poorly understood. Our goal is to identify features of the human microbiome that predict microbial community stability and resilience following disturbance. We propose an innovative large-scale clinical study design that will generate the necessary compositional and functional data from the most relevant ecosystem, i.e., humans! We will develop novel statistical and mathematical methods for data integration (sparse, non-linear multi-table methods), and test existing ecological theories and apply statistical learning strategies to allow data-driven investigation of ecological and clinical properties that determine and predict stability and/or resilience. The breadth and magnitude of this project's impact are significant: We envision tests to predict microbial community responses to disturbance, and procedures to stabilize or restore beneficial microbial interactions as needed. A predictive understanding of the stability and resilience of the gut microbiota will advance the rational practice of medicine. There are three key innovative aspects to our approach: 1) sequential perturbations of different types in a large number of human subjects sampled over time;2) multiple compositional and functional measurements made on the same samples;and 3) novel data integration methods that incorporate all of the information.
Aim 1. Profile the human microbiome before, during and after multiple forms of disturbance. One hundred subjects will each be sampled at 40 time points over a 34 week study period that encompasses two types of perturbation in each subject (dietary shift, and bowel cleansing or antibiotic). From each sample, we will determine taxonomic composition, genomic content, meta-transcriptome, and metabolomic profiles.
Aim 2. Discover resilience: Develop non-linear approaches for complex data integration using sparse, multiple-table methods. We will develop a novel sparse, multiple-table approach for data integration and simultaneous analysis of diverse types of complex data over time.
Aim 3. Explain resilience: Use statistical learning approaches to find the predictive features that characterize resilience. Using the multiple table approach, we will compare routine unperturbed dynamics within a community to the varied responses to a perturbation, define stable states, and identify common network features characteristic of resilient communities subjected to different forms of disturbance. Finally, we wil use validation techniques to confirm these candidate predictors of community resilience.
Humans rely on the microbial communities that colonize the gut for a wide variety of critical functions, including nutrition, immune system maturation, protection against infection by disease-causing microbes, and detoxification of environmental chemicals. Daily life is punctuated by events, such as exposure to antibiotics or other chemicals, or changes in diet, that sometimes disturb or destabilize our microbial communities with potentially severe and sustained negative impacts on health. We propose an ambitious study in which we will monitor the microbial communities of healthy humans before, during and after several types of planned disturbance, and discover community features that predict future stability or future recovery from disturbance, with the expectation that our findings will fundamentally change the practice of medicine.
|Nguyen, Lan Huong; Holmes, Susan (2017) Bayesian Unidimensional Scaling for visualizing uncertainty in high dimensional datasets with latent ordering of observations. BMC Bioinformatics 18:394|
|Callahan, Benjamin; Proctor, Diana; Relman, David et al. (2016) REPRODUCIBLE RESEARCH WORKFLOW IN R FOR THE ANALYSIS OF PERSONALIZED HUMAN MICROBIOME DATA. Pac Symp Biocomput 21:183-94|
|Josse, Julie; Holmes, Susan (2016) Measuring multivariate association and beyond. Stat Surv 10:132-167|
|Callahan, Benjamin J; McMurdie, Paul J; Rosen, Michael J et al. (2016) DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 13:581-3|
|Callahan, Ben J; Sankaran, Kris; Fukuyama, Julia A et al. (2016) Bioconductor Workflow for Microbiome Data Analysis: from raw reads to community analyses. F1000Res 5:1492|
|Bik, Elisabeth M; Relman, David A (2014) Unrest at home: diarrheal disease and microbiota disturbance. Genome Biol 15:120|
|Modi, Sheetal R; Collins, James J; Relman, David A (2014) Antibiotics and the gut microbiota. J Clin Invest 124:4212-8|