Oh this is awesome. I always wished there would be something like this along the lines of blood tests. Keep an anonymous public database of blood tests results with known diagnosis of the patient, age, ethnicity, global region, etc and publish it out. Can be completely anonymous and people would basically just get an immutable entry every time they have a blood test over their lifetime with whatever the dataset happened to be that day.
Would be an excellent way to mine for commonalities among different conditions. I just assumed there was already something like that, but private. Seems like it would enable fairly significant study.
Why don't they have a public, encrypted database where anyone can use their private key to sign certificates authorizing providers to post information about them, and the providers sign it with their certificates also? So you can share (some of) your health history with anyone you want to, but otherwise it would be anonymous and private??
I've often asked your first question, and I think the answer is because it's just not simple for an ordinary person. The public benefit is obvious, but it'd have to be decentralized and the technology for that just isn't easy enough for most people to use.
Due to the nature of genetic information, it can never be really anonymous if you are genotyping genome size amounts of host DNA. There is simply too much data to anonymize it's provenance.
Microbiome is generally anonymous because you can filter out host DNA from it. Though, I bet you could still trace it back to an individual if you tried.
From http://americangut.org/our-results-so-far/ - "The American Gut project has many more samples representing more groups of people than other studies, such as the Human Microbiome Project, Global Gut, or Personal Genome Project."
No hard numbers on their site, but there's over 7500 contributors on their sample collection kit fundraising page.
This is excellent and can only help to improve the (currently dismal) understanding of the gut biome.
But, it only measures bacteria and archaea. To eukaryotes and fungi, it is blind. And those are obviously I'm important components of gut flora.
Since they can be measured fairly inexpensively with more conventional means, perhaps it would improve the science if they could charge double the $99, and get a basic correlated measure of common eukaryotes and fungi in each sample.
And viruses, too! I am very curious to find out what sorts of bacteriophages are part of the picture. In general we have some sense that actively changing the ratios of different bacterial species (e.g. with antibiotics, or fecal transplants) can address some health problems - but it's possible that a targeted phage might be another way to accomplish that as well. (edited for fuller thought)
"All of the microbial sequencing data and health and lifestyle information that we collect are made public through the European Bioinformatics Institute, enabling researchers from all over the world to ask groundbreaking questions using the American Gut dataset."
Yes. They even provide the Jupyter notebooks that power all their analyses so you can re-process and re-analyze the data. I was looking into AGP recently because I had the idea that variance component estimation could be used to quantify how much gut bacteria composition matters, along the lines of GCTA for genes, and they were by far the most open and best dataset around.
Even though it seems microbiome is a science buzzword, it might be our entry to understanding, quantifying, the individual biologic differences between humans.
I long for the day I can take a pill that analyzes your gut and tells you what are the effects of your diet, and what you could do to improve ______ (concentration, tiredness, excrements, muscle buildup).
The term "microbiome" has become tired and 'frothy'. I advocate using the unglamorous but old school term "microbial flora" however futile and unimportant my effort may be.
How about a smart toilet rather than a pill for the daily diet analysis?
Would be an excellent way to mine for commonalities among different conditions. I just assumed there was already something like that, but private. Seems like it would enable fairly significant study.