A couple of months ago, I called attention to a concerning matter: a science writer, Tina Saey, reported on Twitter that she received dramatically different results from the two major microbiome sequencing services available, uBiome and American Gut. And what was most concerning was that the two tests were done using the same sample. So any sort of external factors like diet, or simply the passage of time, could not account for the difference. Tina later posted a lengthier rundown and analysis of her experience.
Shortly thereafter, another science writer undertook a similar experiment, but with a twist: like Tina, she sent the same sample to both uBiome and American Gut, but she also sent an extra sample to American Gut that was taken from the same — but a different section of — her stool (sorry, I try my best to avoid the inevitable ick factor). Interestingly, in her case both uBiome and American Gut matched up well. But her two American Gut samples, taken from different parts of her stool, did not. There were a couple of fairly significant discrepancies.
Well, I think it’s time to close the loop on this matter as best we can, as I’ve been getting questions about the usefulness of testing.
A couple of weeks after sounding the alarm here on this blog, uBiome co-founder Zac Apte left a comment here directing people to an informative blog post at uBiome addressing this issue. The main thrust of it is that uBiome believes the discrepancy to likely be due to the fact that they and American Gut use two differing “extraction” techniques, which are known to result in different outcomes. Fortunately, the way these techniques produce different outcomes is well-understood and can be quantified, and so Zac and his team were able to run an analysis to see if Tina’s results matched when corrected using the “extraction bias” information. And they did seem to come quite close:
Furthermore, I queried Zac recently on the case of that second writer and her experiment on intra-stool variation. And his answer was that this is not much of a surprise — intra-stool variation is expected, and that is why these services have specific collection instructions (swabbing toilet paper), so that the same area is sampled every time. So it seems that directly sampling stool, and in two different areas, circumvents the procedure that’s specifically designed to avoid this problem.
So where does this all leave us?
Well, we’re certainly still left with the question of which technique — and thus which service — provides a more useful and “accurate” accounting of your gut microbiota. Ideally, I’d like to see a dialogue between American Gut and uBiome so that we can get a better understanding of all this and get at what’s really going on. A comparison study might also reveal that perhaps something other than the extraction technique is creating the discrepancy, such as differences in transport or collection methods.
But regardless of all that, I think there are two strong reasons to do one of these tests if you are so inclined.
In my opinion, this is the clearest and best reason to plunk down the money for these tests: the chance that you will find something weird and out of the ordinary. Who knows what you might find? That’s really the usefulness of a test that is largely based on comparative data to a broad population. And it seems to me that both services are still capable of surfacing this kind of thing. What do I mean by something weird? Something like an overgrowth of a specific class of bacteria, the presence of something very rare, or bacterial ratios that are significantly outside the distribution of the population to which you belong.
Detection of any of these things are helpful clues that may get you closer to figuring something out. If you suffer from some sort of health issue and suspect a gut bacteria-related connection, and you can afford the cost, then this may very well be a good reason to do this kind of testing.
Trending — seeing how your gut bacteria change over time in response to different stimuli and interventions (drugs, diet, etc.) — is the other very good reason to consider testing. Assuming you stick with one company for your testing, you should be able to get a pretty reliable view of how your gut is changing over the course of several tests. Trending might also be useful for trying to connect the improvement or deterioration of a specific health condition to a specific change in your gut bacteria. Also, while we don’t have a ton of clear and widely established bacterial health markers, the research definitely does point to some (and if you’ve been following along, you should know who they are). Which means there are some clear targets for anyone trying to hack their gut.
The one piece of advice I would give in this regard is to always try to make a prediction before testing, instead of trying to make connections in hindsight. It’s very easy to make all sorts of connections and explanations after the fact, but if a prediction made before receiving results turns out to be true, it’s much more likely to be relevant (because it’s so unlikely for it to actually come true).
And on that final note: I’m going to have some pretty interesting stuff to share very shortly. Stay tuned.