Mapping anti-microbial resistance with metagenomics methods


I am sitting at a conference in Istanbul, looking at a fantastic project called METASUB. I am no biologist, so bear with me and correct any mistakes I might make. I am sharing it because I think it might be a fun thing to think about, and maybe even work on, for people like @Dorsaf, @amiridina, @JasperB, @ramykim, @rachel etc. I can specifically see a potentially marketable project in northern Africa. Here goes:

  • The metabiome of a place is the microbial ecology of its whole human population. METASUB collects swabs in places like metro stations, sewage treatment systems and public beaches.
  • It turns out you can find detect the existence of superbugs (resistant to anti-microbial treatment) by looking for certain genes, ignoring the bugs themselves.
  • You can then cross the genes signaling superbugs with the existing antibiotics that those superbugs are resistant to. This will tell doctors in the city not to use those antibiotics. The final product is a matrix: common bugs on the Y, antibiotics on the X. The cell (A, B) takes value 1 if bug B is resistant to antibiotic A. If column A is full of ones, don’t use A in medical practice.

What do you guys think?


Makes sense! I wonder how useful that method is, given that there are (I think) big differences between place. Hospitals, cattle farms, fish farms, … are potential hotspots of superbugs. Subway stations not so much. We already kind of know where to look for them. Expanding where we look for them would decrease the effectiveness, unless it somehow enables us to find new hotspots we don’t know about.

I’m no expert, just some thoughts off the top of my head. Waiting for a real expert to explain why what I just wrote is wrong :slight_smile:


I wonder about the data found from the subways too. Sometimes you find whatever you look for, but it is not really relevant to who gets infected. (i.e. legionella bacteria can supposedly be ‘found’ everywhere with metagenomic methods, but the outbreaks only occur in a limited subset with particular susceptibilities…)


one more point - another ‘for instance’ - for the microbiome, what you can readily culture is not the main inhabitant, apparently, from ubiome etc info…