6.4: Study 3- Human Gut Ecology (HuGE) project
- Page ID
- 40946
This study aims to identify more than three hundred dietary and environmental factors affecting human microbiome. The factors, which were regularly tracked by an iPhone App, were the food the subject ate, how much they slept, the mood they were in etc. Moreover, stool samples were taken from the subjects every day for a year in order to perform sequence analysis of the bacterial group abundances for a specific day
Figure 6.3: Abundance levels of different bacterial groups in control patients, Crohn’s disease patients and patients with ulcerative colitis.
relevant to a particular environmental factor. The motivation behind carrying out this study is that, it is usually very hard to get a strong signal between bacterial abundances and disease status. Exploring dietary effects on human microbiome might potentially elucidate some of these confounding factors in bacterial abundance analysis. However, this study analyzed dietary and environmental factors on only two subjects’ gut ecosystems; inferring statistically significant correlations with environmental factors would require large cohorts of subjects.
Figure 6.4 shows abundance levels of different bacterial groups in the gut of the two donors throughout the experiment. One key point to notice is that within an individual, the bacterial abundance is very similar through time. However, bacterial group abundances in the gut significantly differ from person to person.
One statistically significant dietary factor that was discovered as a predictive marker for bacterial popu- lation abundances is fiber consumption. It was inferred that fiber consumption is highly correlated with the abundance of bacterial groups such as Lachnospiraceae, Bifidobacteria, and Ruminococcaceae. In Donor B, 10g increase in fiber consumption increased the overall abundance of these bacterial groups by 11%.
In Figure 6.6 and Figure 6.7, a horizon plot of the two donors B and A are displayed respectively. A legend to read these horizon plots is given in Figure 6.5. For each bacterial group the abundance-time graph is displayed with different colors for different abundance layers, segments of different layers are collapsed into the height of a single layer displaying only the color with the highest absolute value difference from the normal abundance, and finally the negative peaks are switched to positive peaks preserving their original color.
In Figure 6.6, we see that during the donor’s trip to Thailand, there is a significant change in his gut bacterial ecosystem. A large number of bacterial groups disappear (shown on the lower half of the horizon plot) as soon as the donor starts living in Thailand. And as soon as the donor returns to U.S., the abundance levels of these bacterial groups quickly return back to their normal levels. Moreover, some bacterial groups that are normally considered to be pathogens (first 8 groups shown on top) appears in the donor’s ecosystem almost as soon as the donor moves to Thailand and mostly disappears when he returns back to United States. This indicates that environmental factors (such as location) can cause major changes in our gut ecosystem while the environmental factor is present but can disappear after the factor is removed.
Figures from the David lab removed due to copyright restrictions.
Figure 6.4: Gut bacterial abundances plotted through time for the two donors participating in HuGE project.
Figures from the David lab removed due to copyright restrictions.
Figure 6.5: Description of how to read a horizon plot.
Figures from the David lab removed due to copyright restrictions.
Figure 6.6: Horizon plot of Donor B in HuGE study.
In Figure 6.7, we see that after the donor is infected with salmonella, a significant portion of his gut ecosystem is replaced by other bacterial groups. A large number of bacterial groups permanently disappear during the infection and other bacterial groups replace their ecological niches. In other words, the introduc- tion of a new environmental factor takes the bacterial ecosystem in the donor’s gut from one equilibrium point to a completely different one. Even though the bacterial population mostly consists of salmonella during the infection, before and after the infection the bacterial count stays more or less the same. The scenario that happened here is that salmonella drove some bacterial groups to extinction in the gut and similar bacterial groups took over their empty ecological niches.
In Figure 6.8, p-values are displayed for day-to-day bacterial abundance correlation levels for Donor A and B. In Donor A’s correlation matrix, there is high correlation within the time interval a corresponding to pre-infection and within the time interval b corresponding to post-infection. However, between a and b there is almost no correlation at all. On the other hand, in the correlation matrix of donor B, we see that pre-Thailand and post-Thailand time intervals, c, have high correlation within and between themselves. However, the interval d that correspond to the time period of Donor B’s trip to Thailand, we see relatively little correlation to c. This suggests that the perturbations in the bacterial ecosystem of Donor B wasn’t enough to cause a permanent shift of the abundance equilibrium as in the case with Donor A due to salmonella infection.
Figures from the David lab removed due to copyright restrictions.
Figure 6.7: Horizon plot of Donor A in HuGE study.
Figure 6.8: Day-to-day bacterial abundance correlation matrices of Donor A and Donor B.