Human nature likes simple answers with simpler solutions. If you state that SIBO is caused by an overgrowth of Enterococcus faecalis and that berberine kills this bacteria – a lot of people will take it as gospel even when faced with dozens of studies listing over a dozen different bacteria involved.
Many people using my site are looking for simple, easy, effortless analysis and resolutions. The reality is that we are dealing with an extremely complex and interacting system that is constantly changing.
Your microbiome is constantly changing — usually at a slow rate. The association of microbiome dysfunction with symptoms and medical conditions is getting stronger and stronger. It has already open treatment options for some almost untreatable conditions.
Purpose of Microbiome Comparison Pages
These pages purpose are two fold:
- To more easily identify what is changing as a result of probiotics, diet changes, and even work stress.
- To try to identify which of thousands of bacterias may be responding to these changes. If the changes are bad, then we want to know how to change things.
The microbiome comparison pages have gotten a face lift. They give more data and are easier to use. If you have two or more samples uploaded, you will see the button shown below on the Available Samples page
Clicking it will take you to a new page showing some high level statistics. In general, the lower these numbers, the better you are. These are the high level simplification into just 4 numbers for each of your microbiome samples.
How many “Match Taxonomy” should you expect?
The chart below shows what is in the uploaded sample. Some key statistics:
- Average is 142 (It is 146 for those deeming themself to be healthy)
- Median is 147
- Bottom 10% is 114 or less
- Top 10% is 167 or more
This uses the data from our Eureka discover that some symptoms have a statistically significant match to some bacteria. See this post for more information. For some 200 bacterias that matched to some combination of bacteria we do the following:
- Count how many of these bacteria that you have (Match Taxonomy)
- Count how often your values matches the pattern discovered, then convert it to a percent.
For a totally representative microbiome, we would expect to see 25%. We used 4 “bins” to get the Eureka data. We put the same number of samples for a specific bacteria in each these bins. This means that we would expect to have 25% of our bacteria match up if things are random.
We see 75% on one sample — Think of a four sided die that comes up 75% of the time — does that sound like a fair die? No. Neither does this sound like a healthy microbiome.
A second dimension of this measure is the number of matches. Above we see as low as 136, below, another person has 173. This is a proxy for measuring microbiome diversity — the higher your count, the more diverse you are.
From medical literature studies we have gathered a list of bacteria reported to statistically significant over or under representation in some study. We do not know what would be expected on a healthy population, nor can we compute odds. Comparing independent microbiome studies is very difficult to do. These numbers are a general feel set of numbers. A downward movement of the numbers is usually deem to be good — you are moving closer to “control” values.
My Own History
The example above is from a reader that showed some contrasts. Let us look at mine since I was in remission for 2 samples and then have a flare that is on-going (but improving) – with 3 samples (a 4th was mailed today)
The first two numbers above were my being in remission from ME/CFS. The first set showed increasing microbiome diversity and less symptom matches (good); the second showed more matches to conditions (bad). I would deem the change between the first two as being inconclusive.
With the flare, we had dramatic changes: Diversity dropped and symptom matches increased (bad). On the condition front, it seems like we are improving but the change is very dramatic (bad).
Looking at the three flare microbiome samples, we see the second one being the worst with the third one showing sign of recovery (i.e. moving back closer to the pattern seen during remission.
- At onset, the number of bacteria shifting towards symptom association increased by almost 10%.
- At the same time, various low bacteria associated with conditions increased considerably.
- Lots of changing symptoms
- At the 2nd sample, even more symptoms hits occurred
- We had the appearance of instability — high bacterias dropped by 80% and low ones started to increase.
- At the third sample, we seem to be heading back to restoration of the former (pre-flare) values:
- A 11% drop in bacteria matches suggests that bio-diversity was taking a hit.
- Condition Bacteria continue their restoration to pre-flare levels.
All of this comes out of 4 numbers for each sample.
This page compares your values to those seen with statistical significance. Remember, all of the samples are divided into 4 bins with the same number of samples out in each. If one bin has many excessive samples for a certain condition, it may be marked as such (after testing for statistical significance). This page shows which bacteria you have matches for — you are expected to have some by random chance (25% of the total) – in this case 160/4 = 40. So we can not say any item is significant by itself.
- Very High: Match the top 25%
- High shift: above median but not in very high
- Low Shift: below median but not in very low
- Very Low: Match the bottom 25%
My suggestion would be to start crossing out Low-Shift and High-Shift until you get the expected number (40) – working downwards from Phylus -> Class -> Family -> Genus (I may mechanize that in a future update). What is left over are likely the main bacteria that contributes to health issues.
This compares against templates for various conditions based on multiple studies. We see a major jump in bacteria associated with various conditions in the first microbiome done during the flare – i.e. almost 50% increase of bacteria with a high association, this quieted down in subsequent weeks – but these overgrowths likely contributed to the flare.
This is effectively a symptom tracker. The usefulness depends on the accuracy of information entered. It also allows us to do citizen science.
This done old school — using averages. It may be revised in the future to use BoxPlot methods (which should quiet down the noise).
If you have entered metabolism data from ubiome, then you can see how these functions have altered. The terms are those used by Ubiome and you may need to google to understand them and what they mean.
If you want to know how to enter this data, look at this post (which now has a rough video included).
Everyone is different and there can be no one path fits all formula. We can provide information of what changes — and potentially offer suggestions of how to correct things. These are suggestions only — because we do not know, nor can we expect to know in the next century.