An update on results of prior analysis

This is a follow up of several prior posts

This person did his tests using and then transfer the data to

I did the special studies for a couple months taking symbio, florastor, GOS,Arabox., d ribose, pea fiber, etc. 

Feel in general more energized, specially after the round of florastor which I had done just a four days then tested at the time so impact probably won’t fully show in this test. 
Where to go from here? Drop special studies focus  or stay the course?

Why Follow Up Posts are important

The first item is simple, does the model and suggestion appear to work. Everything is theoretically computed, not based on clinical practice or clinical studies. The second item is that these posts encourages people to try suggestions, or to do “self-serve” with the site.

Foreword – and Reminder

I am not a licensed medical professional and there are strict laws where I live about “appearing to practice medicine”.  I am safe when it is “academic models” and I keep to the language of science, especially statistics. I am not safe when the explanations have possible overtones of advising a patient instead of presenting data to be evaluated by a medical professional before implementing.

I cannot tell people what they should take or not take. I can inform people items that appears to have better odds of improving their microbiome as a results on numeric calculations. I am a trained experienced statistician with appropriate degrees and professional memberships. All suggestions should be reviewed by your medical professional before starting.

Comparisons between Samples

We will start by adding new columns for the latest sample. The person measured with Ombre and then transferred data to Biomesight to get a second interpretation of the raw digital data.

CriteriaBS 6/6BS 7/19BS 11/22OL 6/6O 7/19O 11/22
Lab Read Quality2.
Bacteria Reported By Lab280497468365628473
Bacteria Over 99%ile270561
Bacteria Over 95%ile2431227249
Bacteria Over 90%ile495812495121
Bacteria Under 10%ile1762187186087
Bacteria Under 5%ile53090102846
Bacteria Under 1%ile01322172
Rarely Seen 1%040090
Rarely Seen 5%418158409
Outside Range from JasonH449729
Outside Range from Medivere171720161620
Outside Range from Metagenomics777777
Outside Range from MyBioma9910141410
Outside Range from Nirvana/CosmosId222222232322
Outside Range from XenoGene6610111110
Outside Lab Range (+/- 1.96SD)613210142
Outside Box-Plot-Whiskers708429646125
Outside Kaltoft-Moldrup701137011218291
Condition Est. Over 99%ile110000
Condition Est. Over 95%ile240000
Condition Est. Over 90%ile560220
Enzymes Over 99%ile310113155
Enzymes Over 95%ile4632166982147
Enzymes Over 90%ile9051103155411405
Enzymes Under 10%ile10221935455138169
Enzymes Under 5%ile45132154226778
Enzymes Under 1%ile64729520
Compounds Over 99%ile972910412619
Compounds Over 95%ile567623338539789
Compounds Over 90%ile292313347533548118
Compounds Under 10%ile72125264183248133
Compounds Under 5%ile3964163109127100
Compounds Under 1%ile52141161742
Note: I just cut and pasted from “Multiple Samples” tab to Excel to make the above table.

What do I see above?

  • Sample Quality are the same (expected from using the same FASTQ file)
  • Rare and very high bacteria have a significant improvement
  • All of the statistical out-of-range measures(Std Dev, Box-Plot, K/M) reduced the count significantly.
  • Most of the expert suggested ranges increased.
  • Condition profiles dropped to zero.
  • We see the fragileness of some measures to the software being used to interpret the raw data.
    • Enzymes Overone dropped from prior and the other increased from the prior

My general opinion is that the person has improved objectively. The algorithm explicit goal is to reduce all of the statistical out-of-range measures. Ideally, it will also “fix” the person but that is more complex, we lack sufficient knowledge to hand pick the bacteria. We can get associations to specific bacteria — association is NOT causality often (despite many politicians claiming such!).

Going Forward

KEGG Computed Probiotics

Both labs resulted in the same priority: Escherichia coli at the top, the soil based mixtures, then Bacillus subtilis (I personally prefer to get it “au natural”, i.e. in the traditional Japanese Soy based food, Natto).

BBC Japan’s most polarising superfood?

Special Studies Numbers

I am going to skip them, mainly because the results are erratic until I get a better understanding of this. See Caution: Special Studies Suggestions.

There is a possibility of both being right. Right meaning shifting from the current dysfunctional equilibrium. This could be visualize as shown below. I am seeking understanding and building different approaches. Each approach could work for some (but not others). Too many factors for certainity.

Going forward

I am building a consensus report from the items marked 2 above using the Special Studies. The list is similar to other people with ME/CFS. We see 2 E.Coli probiotics (symbioflor 2 e.coli probiotics, colinfant e.coli probiotics) at the top with d-ribose (a sugar used by E.Coli). This is then followed by the earth based probiotics( General Biotics Equilibrium, Prescript Assist (Original Formula), Prescript Assist (2018 Formula)).

For Consensus building, I am going to use the three statistical selectors with both Ombre and BiomeSight:

Then we will do an uber-consensus (combining the consensus from each)

The suggestions look very typical for ME/CFS persons. I attached the Simplified Suggestions

The probiotics suggested are

  • akkermansia muciniphila
  • bacillus clausii
  • bifidobacterium (animalis)lactis
  • bifidobacterium breve
  • enterococcus faecium
  • lactobacillus brevis
  • lactobacillus bulgaricus
  • lactobacillus casei
  • lactobacillus paracasei
  • lactobacillus rhamnosus
  • lactobacillus salivarius
  • pediococcus acidilactici

I have marked the lactobacillus to indicate that they do not play well with E.Coli probiotics. Thus, should be in an alternative probiotic cycle.

We see by individual fibers, that the separate suggestion of a low fiber diet

Fiber suggestions

The New Boy on the Block: Food Suggestions

See New Suggestions Page — Food Centric. This was trying to mine the available data more to get practical suggestions.

  • Flavonoids: Only Barley was a positive (for Ombre) , it was not on Biomesight list
  • Food Contents: Again, only Ombre had positive suggestions: Brazil Nuts and Olive Oil. Biomesight data disagreed on the Brazil Nuts

Why so few? Why labs contradict? This comes down to two key challenges: Different interpretation of the bacteria from the digital data; a low volume of studies on the substances we are using to build food suggestions. Also, the suggestions are based on some of the contents of the food; there are other parts of the food that will have other effects. This is why direct food, herb and spice studies are best. Every food is a complex mixture of chemicals. Some may help, some may hurt. Care must be taken to avoid the simplistic logic that “Super Breakfast Food contains barley, thus it is good/healthy to eat!” Ignoring the 10 grams of sugar in this product.

Thus, these suggestions should be taken with a grain of salt. They are better than random choices, but far short of what we would ideally like.

Bottom Line

I ran some of the Special Studies suggestions and did a download of simplified consensus. Between approaches, we had agreement on taking:

  • Probiotics:
    • akkermansia muciniphila
    • bifidobacterium (animalis)lactis
    • lactobacillus salivarius
    • saccharomyces boulardii 
  • Other
    • Vitamin K2
    • Calcium
    • Echinacea
    • Omega-3 fatty acids
    • Pomegranate
    • Rutin
    • Tea Tree oil

As well as agreement on avoiding

  • fructooligosaccharides (FOS)
  • jerusalem artichoke
  • Flaxseed
  • Vitamin B2 Riboflavin

“This is too complicated” is what I can hear some people saying. This analysis digs into the nature of the data which is really not needed for most people. I am trying to get better understanding. It looks at some of alternative methods of getting suggestions. It is likely of interest to those treating microbiome dysfunctions as it illustrates many of the challenges in interpreting.

For most people, the best process stays the same:

Why is consensus important? Simple, we have very incomplete data and also have limited accuracy with the microbiome tests. Going the consensus approach is similar to using a Monte Carlo Simulation, an appropriate approach to deal with complex processes with many parameters that are fuzzy that produces better results.