Detecting what is wrong

There are actually multiple paths that will result in different collections of abnormal bacteria is a microbiome. There is no clearly better than the others.

  • Outlier from a reference control population which has been used by the lab
from post: A US Microbiome Test: GI-MAP
  • High or Low from a general population
    • This is seen with ubiome and other 16s reports. An arbitrary value is applied for being High or Low (i.e. 2x or 0.5x)
    • This was the original method of making predictions on the site. It is available on the Advance Analysis page.
example from ubiome
For Fatigue after exercise, we check to see if your values fall in the top or bottom 25% shown above. If so, we would deem it to be abnormal because of your symptom.

Once you identify the outliers — how do you correct?

All of the suggestions on the site are done the same way:

  • Eliminate all bacteria that are not deemed abnormal
  • Give a positive number for bacteria that are too high, we call this weight
  • Give a negative number for bacteria that are too low,
    we call this weight
  • Scan a list of known modifiers (which have a value of 1 if it increases the bacteria, -1 if it decrease the number) multiplying by the weight
  • Sum up each modifier and then order them by whether the net sum improves or deteriorates this collection of bacteria as a whole.

There are a few variations, for example, taking the log of the weight to moderate the weight of more extreme values.

But, but, but…. I have no …

Some populations with have none of specific bacteria because other bacteria serve the same function. DNA and diet selected the best providers.

If you believe that you need certain bacteria and you have none. then nothing may appear on the suggestions. What appears on suggestions may be items that eliminates the bacteria that kills these certain bacteria (thus allowing them to recover).

You can look up directly what impacts these bacteria under the Reference Tab on the site. The most common requests are shown below: