Biomesight vs Thorne Tests – Differences

This is a person as in this prior set of posts, A Microbiome Trek Continues thru the land of ME/CFS.

There are two key differences that needs to be understood

  • Difference in numbers reported (of bacteria and percentiles)
  • Will the suggestions change?

Comparison of Results: Thorne to Biomesight

The samples were only a few weeks apart, so similar data. The ratio of bacteria reporting is 6x more for Thorne than Biomesight, so the expectation would be similar shifts for most of the others.

Key Difference: Bacteria Percentiles come from Thorne, the other percentiles are computed against a composite of other samples (until we get sufficient samples). The Conditions, Enzymes and Compound estimates are likely unreliable (we compare against all tests and not other samples from the same procession) and we will ignore in this analysis.

Many of the criteria are identical between tests: Outside Range for JasonH, Medivere, Metagenomics , MyBioma, Nirvana/CosmosId and XenoGene. So for people using those criteria — there is no difference between the tests.

The Bacteria over 90% and Bacteria under 10% are a simple statistic to understand. 10% should be under 10% and 10% above the 90%ile to have a balance microbiome.

With Thorne we have 3226 bacteria and true randomness then you would expect around 322 in each group. We find 239 over 90%, close, but a whopping 1577 under 10% — that 48% of all bacteria, not 10%!!! In other words, we have a massive number of different bacteria at low levels. It is not a problem of a few bacteria being too high (which is a common belief about gut dysfunction), but many only have token amounts.

For Biomesight, we have 503, and thus would expect 50 and 50. For over 90%ile, we have just 25, and for under 10%, 108 bacteria. The high %ile is just 50% of expected and 200% of expected for low with Biomesight; Thorne is just 75% of expected for high, but a massive 489% of expected for low.

CriteriaThorne SampleBiomesight
Bacteria Reported By Lab3226503
Bacteria Over 99%ile19310
Bacteria Over 95%ile21218
Bacteria Over 90%ile23925
Bacteria Under 10%ile1577108
Bacteria Under 5%ile141144
Bacteria Under 1%ile11063
Different Labs – Items Skipped
Outside Range from JasonH66
Outside Range from Medivere1616
Outside Range from Metagenomics77
Outside Range from MyBioma55
Outside Range from Nirvana/CosmosId1717
Outside Range from XenoGene3535
Outside Lab Range (+/- 1.96SD)1898
Outside Box-Plot-Whiskers68527
Outside Kaltoft-Moldrup175391
Condition Est. Over 99%ile66
Condition Est. Over 95%ile157
Condition Est. Over 90%ile2410
Enzymes Over 99%ile9310
Enzymes Over 95%ile673118
Enzymes Over 90%ile1131647
Enzymes Under 10%ile312150
Enzymes Under 5%ile26275
Enzymes Under 1%ile18312
Compounds Over 99%ile230100
Compounds Over 95%ile498463
Compounds Over 90%ile684606
Compounds Under 10%ile1350599
Compounds Under 5%ile1336580
Compounds Under 1%ile1324569

Comparison of Results: Thorne to Ombre

The Bacteria over 90% and Bacteria under 10% are a simple statistic to understand. If you have 3226 bacteria and true randomness then you would expect around 322 in each group.

  • For Ombre we would expect 59 over 90%ile and under 10%ile. close. We have 22 or 37% of expected for low %ile and 117 or 200% of expected for low percentile.

Many of the criteria are identical between tests: Outside Range for JasonH, Medivere, Metagenomics , MyBioma, Nirvana/CosmosId and XenoGene. So for people using those criteria — there is no difference between the tests.

CriteriaThorne SampleOmbre
Bacteria Reported By Lab3226588
Bacteria Over 99%ile1932
Bacteria Over 95%ile21210
Bacteria Over 90%ile23922
Bacteria Under 10%ile1577117
Bacteria Under 5%ile141167
Bacteria Under 1%ile11066
Different Labs – Items Skipped
Outside Range from JasonH77
Outside Range from Medivere1414
Outside Range from Metagenomics55
Outside Range from MyBioma88
Outside Range from Nirvana/CosmosId1818
Outside Range from XenoGene4646
Outside Lab Range (+/- 1.96SD)1895
Outside Box-Plot-Whiskers68534
Outside Kaltoft-Moldrup1753129
Condition Est. Over 99%ile60
Condition Est. Over 95%ile150
Condition Est. Over 90%ile240
Enzymes Over 99%ile930
Enzymes Over 95%ile6739
Enzymes Over 90%ile1131101
Enzymes Under 10%ile312165
Enzymes Under 5%ile26265
Enzymes Under 1%ile1832
Compounds Over 99%ile23038
Compounds Over 95%ile498236
Compounds Over 90%ile684332
Compounds Under 10%ile1350248
Compounds Under 5%ile1336159
Compounds Under 1%ile132433

My personal opinion is that Thorne is better because the more bacteria reported, the greater the statistical significance of over and under representation. On the positive side, all three samples agree on the shifts of bacteria patterns


The distribution continues to match a common pattern with ME/CFS microbiomes, an over abundance of low percentile bacteria. This is also seen with the prior Biomesight sample. This shift is made much stronger with Thorne because more genus and species are reported. It also emphasis the shifts seen above.

0 – 9417628
10 – 198582
20 – 296085
30 – 394269
40 – 493432
50 – 5959339
60 – 692660
70 – 791628
80 – 891531
90 – 9921142
Thorne Report
0 – 92433
10 – 193237
20 – 291523
30 – 39911
40 – 49916
50 – 59814
60 – 6989
70 – 791017
80 – 89715
90 – 99610
Biomesight Report

Treatment Dilemma

The usual algorithm is to increase bacteria with low percentiles and decrease those with high percentiles. When you have a huge numbers of low percentile then the question arises: Do you really want to increase these, or do you want to eliminate them entirely to get them off the radar? It is a valid question, but to do that, we have to make increase/eliminate suggestions on 417 genus. That is not practical (given the sparseness of data and limited knowledge of so many genus). My working hypothesis is that keeping to the usual algorithm is the best way to go. Let the bacteria make the determination of winners and losers.

Going Forward

I am going to build two consensus reports. One for Thorne and one for the latest Biomesight, then use the “Uber Consensus” report on the Multiple Samples tab. The purpose is to see whether there are really significant differences in suggestions between the two sample reports.

We are going to do 4 basic suggestions for each:

The results had 581 suggestions. I did a Pivot tables of Take Counts against Avoid Counts to visualize the similarities between each set of suggestions going into the uber suggestions. I read the pivot table below as indicating that the suggestions were equivalent with 73 items being to Take with no Avoid, and a further 72 items with just 1 avoid. We have lots of choices in agreement


What are some of the top suggestions?

As with the prior reports, Escherichia coli probiotics is at the top of the KEGG computed probiotics, typical for ME/CFS. From the consensus, we have:

The absence of most Lactobacillus is not surprising because they are hostile to Escherichia Coli. My pivot conference report from 1998 had this bacteria being low in ME/CFS patients. With Thorne, we can get actual numbers (16s numbers for Escherichia Coli are questionable). This person Thorne Results is at the 27%ile for Escherichia and 29%ile for Escherichia Coli, which is consistent with that conference report and the KEGG computation for probiotics.

Out of interest, I looked for the %ile on the Thorne results of the consensus suggested probiotics:

Having the actual percentiles for the strains used in probiotics allows us to tune the suggestions. In this case, we should skip any probiotics with bacillus subtilis or clostridium butyricum. There is no point in taking them. On the other side, you have confirmation that the suggested probiotics are likely to have an impact. I give the Thorne results a big 👍 because you can actually determine the probiotics that you likely not benefit from. The 16s results only report a few probiotic species (with questionable accuracy).

The Extras from the Thorne Results

This person did not see the next data on Thorne’s Web Site — but it was in the data CSV file to upload. The virus count with a few having percentiles. The ones without percentiles are rare ones without data.


And Fungi too!

For both of these sets of data, values over 90% should be researched. Fungi are of special concern because they can both be treated often and may also indicate a mold issue around the person.

User Feedback

Thanks Ken! .. and yikes . Mold is my nemesis. I’ve been trying to figure if I’ve had mold / moldy house for years. I did an ermi for the entire house and got a 2 which is really low. Recently I did an air test all around the house and it was pretty low minus a car which I knew was an issue and I’ve been trying to avoid. I’m trying to decipher if it’s a past issue and I can’t detox or current 🧐🧐

Attached is my air test.

Question: yellow highlights on Thorne. Those were 90%+ , there were some non highlighted 90% + results. Should I only research the ones you highlighted?

I should have a new mold urine test result coming too which I’ll send your way once I get it!

Answers: As a general and very rough rule, count the number of items reported (Virus: 35, Fungi:50). Take this number and multiple by a percentile – say 99%ile, and round up. For Virus and Fungi it is 1. This is the number of false positives that would be expected with 99% or higher.

You appear to have an issue. Your mold test fortunately identifies the genus and where it is located. The HVAC and washing machine hints that it is brought into the house, likely on clothes. The attic appears free of the ones that you are high in. I noticed that Malassezia is not reported in the mold test.

Bottom Line

IMHO, getting a Thorne sample is a definite should do at least once. Why? some of your issues may be fungi or virus related. The difference for suggestions of using Thorne, BiomeSight or Ombre is slight. The differences are reasonable given the sparseness of the data that we have for suggestions.

Postscript – 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 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.

The answers above describe my logic and thinking and is not intended to give advice to this person or any one. Always review with your knowledgeable medical professional.

Leave a Reply

Your email address will not be published. Required fields are marked *