What do we hope to accomplish?
Very long answer: A few large studies of the human microbiota have been completed thus far; they have looked at many individuals, but examined only one or a few samples per person. Several temporal studies have looked at dozens or even hundreds of samples per person, but only included a few people. Thus, one major goal of this study is to generate the first dataset from the human gut microbiota that has both an adequate number of samples per person to assess the dynamics of the microbiota (how its composition and activity change over time), and includes enough people so that different dynamic patterns can be recognized, categorized and compared. For example, we might find that gut microbial stability is fairly consistent in people over time, with different people falling on a continuum between high and low levels of stability. Or we might find that particular types of microbes tend to be stable or unstable in their abundance over time, regardless of which person they are found in. Or we might find that some people switch between a small number of stable states, or have periods of unpredictable instability between intervals of stable community structure and function. The number of possible patterns is unlimited; we won’t know what’s typical until we actually generate the data.
A second major goal is to provide a complete dataset across all the samples for each of the analytical technologies we employ, and to develop more effective statistical tools to analyze these complex, high-dimension, heterogeneous datasets. Most studies of the microbiota thus far have used only a single technology (16S rRNA sequencing, metagenomics, or metabolomics), or have included only a small subset of samples when applying multiple techniques. For example, the largest temporal study of any aspect of the human microbiota that has been published (as of 2014) investigated the composition of the vaginal microbiota (using 16S rRNA) with about 1000 samples from 32 women. The researchers also performed metabolomics, but with only 30 of those samples, from only 4 women. With incomplete datasets, it becomes much harder for statistical analyses to detect genuine associations in the data (e.g. that the abundance of particular genes and the concentrations of particular chemicals are correlated over time and across individuals). With few samples, random noise in the data can be mistaken for genuine associations (and vice versa). The same statistical framework we will use to analyze the various types of data obtained from microbiota samples will be extended to include the information obtained about each participant from the questionnaires (and from blood samples and genetic testing for participants who choose to provide them).
The third major goal is to discover factors that explain or predict the dynamic patterns that we observe in the gut microbiota. Factors related to stability and/or resilience are particularly interesting, given their relevance to managing our health – we want a healthy microbiota to be stable, and to be resilient in the face of whatever perturbations come along. On the other hand, we would hope to destabilize an unhealthy microbiota to shift it towards a more healthy state, and we would want to eliminate or at least diminish its resilience, so the unhealthy state wouldn't keep coming back. The ability to compare 3 different types of perturbation (dietary change, antibiotics and colon cleansing), and to compare all of these to routine fluctuations in the microbiota is a powerful feature of our study. We may find that communities differ in their resilience depending on the perturbation, or that resilience is a general property regardless of the type of disturbance. We may find that the pattern of routine day-to-day fluctuations (presumably the cumulative effects of numerous minor perturbations) predicts the resilience of the microbiota in response to larger perturbations. We intend to test predictions made by traditional ecologists about ecosystem stability, such as a link between community diversity and stability.
A second major goal is to provide a complete dataset across all the samples for each of the analytical technologies we employ, and to develop more effective statistical tools to analyze these complex, high-dimension, heterogeneous datasets. Most studies of the microbiota thus far have used only a single technology (16S rRNA sequencing, metagenomics, or metabolomics), or have included only a small subset of samples when applying multiple techniques. For example, the largest temporal study of any aspect of the human microbiota that has been published (as of 2014) investigated the composition of the vaginal microbiota (using 16S rRNA) with about 1000 samples from 32 women. The researchers also performed metabolomics, but with only 30 of those samples, from only 4 women. With incomplete datasets, it becomes much harder for statistical analyses to detect genuine associations in the data (e.g. that the abundance of particular genes and the concentrations of particular chemicals are correlated over time and across individuals). With few samples, random noise in the data can be mistaken for genuine associations (and vice versa). The same statistical framework we will use to analyze the various types of data obtained from microbiota samples will be extended to include the information obtained about each participant from the questionnaires (and from blood samples and genetic testing for participants who choose to provide them).
The third major goal is to discover factors that explain or predict the dynamic patterns that we observe in the gut microbiota. Factors related to stability and/or resilience are particularly interesting, given their relevance to managing our health – we want a healthy microbiota to be stable, and to be resilient in the face of whatever perturbations come along. On the other hand, we would hope to destabilize an unhealthy microbiota to shift it towards a more healthy state, and we would want to eliminate or at least diminish its resilience, so the unhealthy state wouldn't keep coming back. The ability to compare 3 different types of perturbation (dietary change, antibiotics and colon cleansing), and to compare all of these to routine fluctuations in the microbiota is a powerful feature of our study. We may find that communities differ in their resilience depending on the perturbation, or that resilience is a general property regardless of the type of disturbance. We may find that the pattern of routine day-to-day fluctuations (presumably the cumulative effects of numerous minor perturbations) predicts the resilience of the microbiota in response to larger perturbations. We intend to test predictions made by traditional ecologists about ecosystem stability, such as a link between community diversity and stability.