Indoor Microbiome Meta-analysis

A new paper from Adams et al does a meta-analysis of indoor microbiome data to further characterize the microbial interactions within the indoor built environment. They used sequence data from four studies that covered South Korean homes, Colorado restroom surfaces, Colorado kitchen surfaces, and North Carolina homes, and found that communities were most similar between the type of surface versus the type of environment. For instance, toilets in different bathrooms would look more similar than a toilet and sink in the same bathroom. In addition, surfaces showed patterns based on possible human sources. An example given was that plant chloroplast (probably from food) were more common in kitchens than bathrooms. Skin associated microbes were more common in bathrooms than in kitchens, which makes sense considering what the purpose of bathrooms versus kitchens is.

Another cool analysis that was done was running source tracking on the data. This involved defining a “source” like outdoor air, soil, or human skin, then seeing what proportion of a sample’s community is source and what is not. For instance, Oregon classroom air was about 90% outdoor air. Alternately, a NICU in Sacramento had minimal amounts of the defined sources and had a mostly “unknown” source(s).

Using PCoA plots, the authors looked into technical variation and found that samples had a lot of inter-study variation. Other factors that explained the community were location, sample type (toilet vs. chair, etc.), and use of the building.

Another interesting aspect is that they looked at kit controls. Although many people in the field have started using kit controls to examine contamination, not everyone knows what to do with them or how to analyze them. As with previous research done on the subject, this study found that kit controls have their own distinct microbial communities when compared to samples. The authors also found that one taxa that seemed particularly common, Tenericutes, containes bacterial groups that often contaminate cell cultures in the lab as well. There were, of course, taxa that were specific to samples and not common in kit controls. Problematic groups, like human-associated bacteria, that are found in both the kit control and samples pose a problem because it means there’s no way to determine if they are real or contaminants. Removing everything from the samples that is also in the kit could delete real communities that were pre-existing and do not result from contamination.

Overall, this study is fairly thorough in their meta-analysis on the indoor microbiome. I highly suggest reading through the whole paper since there was a lot of information and detail that can’t possibly be covered in a simple blog post.