What’s lurking in the NYC subway? “City-scale metagenomics” brings unprecedented resolution


A remarkable study published last week in Cell Systems has described the ‘environmentome’ of the subway system in NYC. The study has attracted a fair bit of attention in the mainstream press, not least due to claims of Bubonic Plague and Anthrax lurking in the NYC subway system (more on this later)…

The authors took a ‘microbiomic’ approach to characterize the DNA found on surfaces, similar in concept to the Hospital Microbiome Project. Since the collection of microbes in and on our bodies outnumbers our own cells by 10:1, and accounts for up to a third of active molecules in the bloodstream, this sort of approach to microbiology is going to become more and more common.

The study is mind-blowing in many ways, with 1457 samples collected from subways, public parks and alongside a canal. The headline findings are:

  • DNA from pathogens (including Yesinia pestis and B. anthracis and some antibiotic resistant bacteria) were found on some surfaces.
  • Human DNA identified in subway stations mirrors the demographics of the local region.
  • The microbiome of surface in a station is far from stable: hourly sampling over a single day at Penn Station (a busy rail hub) identified considerable ebb and flow in predominant bacterial species.

I have a lingering concern that the techniques available for bioinformatic analysis have not yet caught up with our ability to sequence vast amounts of DNA. Put another way, the identification of bacteria in metagenomic samples is a surpassingly and rather unnervingly approximate science.

In order to identify bacteria in the sample, very many short reads of DNA are produced and then compared with databases. (This results in particular difficulty in distinguishing plasmid from genomic DNA, the subject of a recent post.) Almost half of the DNA identified in the study from New York did not match any known sequences. Furthermore, some of the species identified seem rather unlikely. For example, two of the most common Eukaryotic DNA species identified were the Mountain Pine Beetle and Mediterranean Fruit flies. Now, I’ve spent many-a-weekend in NYC from my time living in Connecticut, and I’ve never seen a Mountain Pine Beetle down there. So, this seems almost certain to be a mis-match with a closely related species due to imperfect databases. And yet when it comes to Y. pestis and B. anthracis, the authors seem more certain that the matches are correct, unlikely as they seem. (There is an acknowledgement in the discussion that these apparent “best hits” may be erroneous.)

Another key limitation is the degree of viability associated with the DNA identified. It could be that much of the DNA is a shadow of ancient contamination that is no longer viable. Whilst the authors did do a small amount of conventional sampling, and did grow some antibiotic resistant bacteria, there is no real sense of how much of the DNA identified is from viable microbes.

Quite a few years ago, I took some similar samples from buses and tube trains in London, and found no MRSA whatsoever and only a few sites grew S. aureus. It would be fascinating to see how a metagenomic analysis of these samples would look. Would there be a vastly different ‘environmentome’ in the London Underground compared with the NYC subway? Probably, but I suspect you wouldn’t find many Mountain Pine Beetles, Y. pestis or B. anthracis in either!

One of the best parts of the study is the accompanying website, which provides an interactive overview of the ‘environmentome’ of NYC: pathomap.org. Related to this, an interesting future application of these data is to derive a persons’ recent or ancient geographical location based on their current microbiome. Criminals beware – analysis of the ‘environmentome’ on the sole of your foot could invalidate your alibi!

The authors should be credited for describing the microbial ecology of the NYC subway in unprecedented detail – and this study will serve as a marker in the sand for future approaches to exploring where we fit into our inanimate environment.

What’s trending in the infection prevention and control literature? HIS 2012 -> HIS 2014

I was privileged to speak at the Healthcare Infection Society meeting in France today on ‘What’s trending in the infection prevention and control literature? HIS 2012 -> HIS 2014’. You can download my slides here, and view the recording below:

I have always enjoyed attending these light-hearted summary sessions at other conferences, so I hope I struck the right tone. In order to track some of the trends in the infection prevention and control literature since the last HIS conference (in late 2012), I plugged some search terms into Google trends (Figure).

Figure: Google Trends for all search terms (excluding viruses) (2004 to present). Logos and arrows represent the time of the HIS 2012 and HIS 2014 conferences. Search terms: hospital cleaning; carbapenem resistant Enterobacteriaceae, whole genome sequencing, fecal microbiota transplantation. [Note, I had to spell it ‘wrong’ (fecal v faecal) to detect a trend. Blasted Americans.]what's trending google trends

Based on my search terms, there was one infection control trend that trumped all others: Ebola. If I include in with the other Google search terms, it eclipses all others! Whilst trends in Google searches may not necessarily correlate with trends in the infection prevention and control literature, in this case, it is true that the outbreak of Ebola in West Africa has prompted a lot of publications in the literature – and consumed an awful lot of professional time for all who are connected with hospital infection prevention and control! Aside from Ebola, other trends in the infection prevention and control literature that I covered include MERS-CoV, universal vs. targeted interventions, faecal microbiota transplantation, whole genome sequencing, carbapenem-resistant Enterobacteriaceae (CRE), and some aspects of environmental science. Finally, I looked into my crystal ball and predict some of the trends in the infection prevention and control literature by the time HIS 2016 comes around.