Did raw potato starch wreck my gut?

What if you were in New York City and somebody gave you a list of the GPS coordinates for every coffee shop in the area. If you knew nothing else, would that information be useful or distracting? The answer depends on what you’re trying to do. If you just want to know, roughly, where people congregate, then it's probably a nice guide. But if you're lost and you need to get across town, then that information may be worse than useless: without knowing the roads or subways, you may end up in the Hudson River.

That’s how I look at my uBiome test results and the wonderful and incredibly detailed analysis done for me by Dr. Grace Liu, who I consider to be one of the world’s best experts about the gut microbiome. Please read her blog, and listen to her Gut Guardians podcast, where she goes into far more detail, but here’s her conclusion about my results:
Unfortunately after amputating over 1/3 of his gut species, many of the phylogenetic core are depleted. The initial levels were awesome but after a high dose of single source of 'fiber', many on re-testing were gone and dramatically diminished numbers.
She’s referring to my “sleep hacking” experiment using raw potato starch to improve my sleep. Another way to put this (in my own words):
Raw potato starch — which is not in any ancestral diet — is bad for health. It may temporarily improve sleep by feeding gut bacteria responsible for the production of most sleep hormones, but it also crowds out other, more important organisms, and opens the door for pathogens that can cause far more trouble.
Sounds pretty dangerous, and of course I stopped all potato starch experimentation immediately after her warning.

Here’s my summary of the uBiome test results that drove her conclusion (pulled straight from my publicly available data)
OrganismMayJunOctRank
Faecalibacterium prausnitzii99571623165790species
Roseburia13554111577825genus
Christensenellaceae825853971340290family
Christensenella269NA38genus
Akkermansia30960196547648genus
Bifidobacteria Longum32NA1858species
Bifidobacterium8473653258747genus
B. Longum as % of total Bifido0.38%NA3.16%
Clostridium350124167971326genus
C. botulinumNA25species
C. clostridioforme289023537215170species
C. baratii1223NA5588species
The units are all uBiome’s “count_norm” field, which you can think of as, roughly, a percentage (a fraction of one million). Items in italics are "good".

Her analysis compares only my May and October samples (she dismisses my Jun result as an anomaly) but what if that's not fair? True, there are some issues with the data for June (for some reason, uBiome computed a much smaller sample that time and you can see from the chart that a few items are missing), but it’s still data, and we can’t be sure that the other samples are any more (or less) accurate. To use the New York coffee shop analogy, when you have almost no information to begin with, are you better or worse off when you drop some data points? The answer is that it depends on what you’re trying to accomplish.

Dr. Liu is alarmed at the drop in some key species between May and October, a fact she attributes to my potato starch experiment. But all of those numbers were already dropping in June. The five months pre-potato starch between May and October coincided with seasonal changes (late Spring to Summer to early Fall), lots of travel, multiple camping trips, and of course the normal dietary shifts that happen as I gained access to the freshly-harvested fruits and vegetables of Summer. Was a week of a couple tablespoons of potato starch really the most important change?

In fact, in this table the only species that reversed course after the initial May-June samples and October were the bifidobacterium (often associated with good health and sleep, and up by a LOT) and the clostridioforme (a potential pathogen, down by a little).

I'm pretty healthy, thankfully, and there are no particular disorders I'd like to treat. Like anyone, I want to feel even better, but in my self-experimentation I certainly don't want to risk falling into some terrible dysbiosis or worse. Potato starch appeared on the surface to help -- the improvement in sleep seemed promising -- but I take Dr. Liu's advice and expertise seriously, so I stopped until I can see more uBiome results. I submitted one sample right after my last experiment, and another a few weeks after that. If potato starch really wrecked my gut, then I'll expect the new samples will show considerable worsening across the board. But if not, then, well, maybe it's okay to continue the experiment. Either way, I'll be taking her bionic fiber advice seriously.

To go back to the New York coffee shop analogy, I think we have to respect how very, very little is known about the terrain around us. When you know absolutely nothing about the critically important gut environment, then a tool like uBiome is such a precious gift of information that it's tempting to use it for much more than it is. We'll need much more data, from many more people, before we can use this information to get across town without falling in the river.