Using Response to Refine Bacteria of Concern

A reader wrote

Another thing I’ve noticed that helps, perhaps 10% of what Amoxicillin helps, is Lauricidin (Monolaurin). I was able to get some sleep last night because I decided to try it.

Monolaurin was not on the top of his recommendations list. It has mixed impact. So the question arises, can we use this response to better identify the bacteria connected with this change of symptoms.

Monolaurin does not come in high, actually low.

Clicking under PubMed we see the bacteria impacted (for better or worse). We have a list (118 in this case). I just added Happy and Sad emoji to clarify if it is a good 😁 or bad 😢 shift

This can still be a long list.

Going over to Citizen Science Special Studies and filtering to sleep issues we find some there – which hints that these may be the key bacteria.

Bottom Line

This is a slow process — for this person, we got clarification quickly.

MCS, Endometriosis, Lyme, Sjogrens, Hypothyroidism etc…

Back Story

  • I took many antibiotics for ear infections as a kid, mainly amoxicillin.
  • At 12 started extreme period cramps.
  • At 18 discovered an ovarian cyst the size of a lemon, then two months later was the size of a grapefruit.
  • Had a c section like surgery to remove it along with 1/3 of the left ovary.
  • After a month of painkillers and whatever else they gave me, I was in severe pain, exhaustion and bowel function stopped.
  • Started rounds of Drs and tests- after a year was diagnosed with FM CFS IBS and within 3 yrs MCS.
  • Was homebound and severely allergic to chemicals for 6 years until I saw a chiropractor in Vegas who did energy work, spinal manipulation, high red meat diet and 6 litres of water/ day.
  • Improvement lasted 4 months and I was able to work but then crashed.
  • I chased it for 10 years, saw many chiros that did this work but never felt as well as the first time.
  • I switched diet again in 2013 to eating more animal protein and veggies and spent 6 months in Costa Rica and was 50 % better.
  • I came back to Montreal and moved into a new place where everything was offgassing and crashed within a week.
  • I took four pills of cipro in 2015 and developed right side pain and sciatic. Since 2018 constant sometimes severe pain in the right leg and hip.
  • I was diagnosed with Sjogrens in 2017 and Hypothyroidism in 2019 and put on cytomel.
  • I have two tiny nodules on my thyroid. I had a tiny cyst in my uterus that burst in 2018.
  • I was diagnosed with Vulvodynia in 2021. It flares in response to certain foods, stress and I think, histamines.
  • They tested me thoroughly for lyme and treated for bartonella with 1 1/2 months of doxy in may 2022.
  • My gynecologist told me last week that I may have endometriosis. I’ve been in perimenopause according to the thyroid doc. I have felt it for a few years now.

My main symptoms are:

Muscle pain, PEM, exhaustion, brain fog, memory issues, constipation ( it is better as long as I stay on top of it), bloating and gas (mostly with fodmaps), gut pain, bad mood, very stressed and angry, very emotional, sleeping issues, crying every day, hopelessness….

Analysis

A long history of microbiome altering events. First Percentages of Percentiles below, which is more extreme than most samples that I have reviewed.

The top predicted symptoms appear to be spot on. Despite all of the other shifts, this appears to persist. A good number of symptoms were correctly predicted.

Going over to health analysis, we actually have a moderate list for General Health Predictors.

Other items:

  • Anti inflammatory Bacteria Score 26.3 %ile
  • Histamine Producers 79.5 %ile — common with ME/CFS
  • Oxalate degrading 0 %ile — suggesting risk of hyperoxaluria or kidney stones.
  • SIBO is reported, which does not appear reflected in the fecal sample.
    • Hydrogen 38.7 %ile
    • Hydrogen sulfide (H2S) 43.3 %ile
    • Methane 4.3 %ile
  • Dr. Jason Hawrelak Recommendations results in 56.4%ile, so in the middle of what is seen in the samples.

PubMed Medical Conditions

None were listed as being significant, so I looked at some of the conditions reported. Remember, having multiple conditions can mask the signature patterns.

  • rosacea: 0 matches
  • endometriosis: 2 of 45 (33%ile)
  • Hypothyroidism 
  • Sjogrens 1 of 35 (23%ile)
  • FM: 1 of 35 (31%ile_
  • CFS: 0 of 64 (0%Ile)
    • ME/CFS with IBS 0 of 18 (0 %ile)
  • IBS: 1 of 68 (8%ile)

So we will include none of these in building suggestions. When there are multiple conditions, patterns are often altered (unfortunately).

Going Forward

The usual “just give me suggestions” (which does 4 different ways of selecting bacteria) plus Special Studies on symptoms

The PDF sections are shown below to give an overview.

Going over to the detail report to address some specific questions. The high was 708 (antibiotics)

Bottom Line

A complex history with a hodgepodge microbiome. Antibiotics occupy the top section of the suggestions — but those are always tricky. “Dr. Knows Best” will often attempt to persuade the patient to take a different one (in complete ignorance of the microbiome impacts).
I would suggest 3 months of the above suggestions and then another microbiome sample to see what has changed.

Questions

Q: My only question is which new feature shows rifaximin as higher up on suggestions?

A: Just enter the name in the consensus (which is now the default screen for “Just Give Me Suggestions”)

Note the stacks of books on the right, it will show the studies that this was based on, in this case 207 citations. There are some 685 bacteria flagged as being atypical, a number of bacteria far higher than usual and indicates severe disruption.

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 can compute items to take, those computations do not provide information on rotations etc.

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.

Analysis Divergence: Biomesight x Microbiome Prescription

In general, I avoid comparing opinions/suggestions from different microbiome resources. Like my earlier The taxonomy nightmare before Christmas… post; some resources may be sufficient/adequate for some people; for others it is not. My criteria for both tends to be simple:

  • More data, and more complete data, tends to better results
    • For the Microbiome it means that Shotgun Analysis where the data is uploadable, complete (often 5000+ items) and has percentile ranking is my preference
    • For the Analysis it means how many substances are considered (MP: 2065), are all interactions considered (MP: 2.5 million), how many different ways of doing analysis are offered (MP: lots).

Whatever you are using may be sufficient. If it is not, then read on.

This person asked for my help on Facebook explicitly and to properly answer, I need to do some comparison of analysis, interpretations and suggestions.

ME/CFS for 9 years. LC from vax injury 2.5 years.

I’ve been following the biomesight recommendations for 18 months and my gut has improved massively. I’ve just completed my third biomesight test and results are in. I have been experimenting with nicotine patches for 6 months now and my fatigue and pem symptoms have improved massively.

However, my most recent results are back and they have never been worse! Do you think Nicotine has a really negative impact on our guts? I can’t explain why everything is soo much worse.

Ken Lassesen / Troy Roach this could be one that you guys could help on.

For info: my gut doesn’t actually feel worse, but the results are terrible 😂.

From a facebook User

Analysis

The reader is relying on BiomeSight evaluation. IMHO there is no single magical number or formula but many features that needs to be examined. Below is a table of the three test results meta-information. Remember that I am use the same measurement of bacteria data as Biomesight.

My general impressions is improvement is continuing despite Biomesight indicating not. Why?

  • Shannon, Simpson and Chao1 Diversity Percentile all moved towards 50%ile from extremes, a good sign.
    • Biomesight Diversity score started at 100% (ideal) and went downwards; completely opposite read to mine.
  • Outside Kaltoft-Moldrup are the ranges that I have the most confidence in, and they continued to drop
Criteria12/7/20232/24/22036/7/2022
Lab Read Quality4.47.310.3
Lab Quality Adjustment Percentage79.789.7100
Outside Range from JasonH677
Outside Range from Medivere131717
Outside Range from Metagenomics799
Outside Range from MyBioma444
Outside Range from Nirvana/CosmosId231717
Outside Range from XenoGene414040
Outside Lab Range (+/- 1.96SD)231624
Outside Box-Plot-Whiskers7961120
Outside Kaltoft-Moldrup53115120
Bacteria Reported By Lab677709866
Bacteria Over 90%ile513286
Bacteria Under 10%ile56263244
Shannon Diversity Index1.7231.9591.914
Simpson Diversity Index0.0680.0460.026
Chao1 Index134681491220849
Shannon Diversity Percentile64.491.887.9
Simpson Diversity Percentile64.643.418.2
Chao1 Percentile72.580.493.8
Lab: BiomeSight
Pathogens373137
Condition Est. Over 90%ile014
  • Biomesight (BS) and Microbiome Prescription (MP) appear to be using different list of pathobionts
    • 7 Dec 2023: MP reported 37, BS cites 49%
    • 24 Feb 2023: MP reported 31, BS cites 72%
    • 7 Jun 2022: MP reported 37, BS cites 36%

The Percentage of Percentiles

The charts are below — we see in the older samples that the 0-9%ile spike that is typical of ME/CFS has disappeared in the latest sample. My preferred single measure of gut health, Chi2 has moved from 60 to 49 to 45. Significant improvement.

Conclusion: Biomesight simple evaluation of overall health may be misleading because it is too simple an algorithm.

Health Analysis

Nicotine Patches Question

Nicotine is one of the modifiers consider by Microbiome Prescription Expert systems.

  • 2022-06-07: Nicotine patch was a low positive
  • 2023-02-24: Nicotine patch was a positive, 5x higher than above
  • 2023-12-07: Nicotine patch was a positive, less than above but 3x the first value.

Suggestions Comparisons

Biomesight just gives suggestions without any attempt to prioritize them. Looking at the suggestions from the latest sample(reader sent the PDF); we list them below. The highest Priority from Microbiome Prescription was 927 and lowest was -906.

Below are Biomesight suggestions followed by how Microbiome Prescription ranks them.

  • Prebiotics
    • Arabinogalactan: Massive Avoid: -906 (based on 331 interactions)
    • Galactooligosaccharides: Avoid -233
    • Guar gum: Avoid -106
    • Gum arabic: Avoid -107
    • Lactose (not in lactose intolerant)
    • Lactulose: Minor take: 81
    • Milk oligosaccharides: Major avoid -233
    • Pectin: Major Avoid: -570
    • Raffinose: Minor avoid: -65
    • Resistant starch: Minor avoid: -26
    • Resveratrol: Avoid -230
    • Stachyose: Avoid – 422
    • Xylooligosaccharides: Avoid: -390
    • Chitooligosaccharides: Minor avoid: -22
    • Yeast beta-glucan: Minor take: 38
    • Psyllium: Minor take: 65
    • Colostrum: Minor Avoid -95
    • Quercetin: Minor take: 39
  • Herbs and Spices
    • Triphala: Minor take: 43
    • Cinnamon: Minor take: 53
    • Ginger: Take: 108
    • Oregano: Take 138
    • Turmeric: Take: 218
    • Thyme: Take: 434
    • Curcumin: Take 180
    • Garlic: Take 200
    • Lauric acid: Take 200
    • Niacin: Major take 732
    • Cranberries: Avoid -22 / -685 (for flour)
    • Olive leaf: Take 179
    • Slippery elm: Major avoid: -803
    • Codonopsis pilosula: Avoid -113
    • Shen Ling Bai Zhu San: Minor take 26
    • high fiber diet: Avoid -99
  • Probiotics
    • Bifidobacterium longum: Avoid -120
    • Lactobacillus rhamnosus: Avoid -471
    • Bacillus coagulans: Avoid -289

So we have a few agreements but a lot of disagreements. It may be just “the change of microbiome environment shock” with either sets of suggestions is causing improvement.

Microbiome Prescription does a holistic approach for suggestions. It looks at the known impact on every bacteria being targeted for a modifier and makes the full details available to review (Click on the 📚). People have been double checking these citations. The decision on Arabinogalactan was based on considering 311 interactions, a few are shown below.

Another difference is that the bacteria selected is based on using 4 different algorithms to select what is of concern and then we do a Monte Carlo simulation on the suggestions.

My impression is that Biomesight considers one bacteria at a time and does not use that many studies to base a recommendation on. I do not know what extent BS consider the complexities of interactions. Biomesight would be the source of information to get better clarity on this.

So what are Microbiome Prescription Top Suggestions

I have placed a 🎯 besides those that are common suggestions

This person has ME/CFS and it is extremely well documented that B-Vitamins moderates those symptoms. Microbiome Prescription shouts out that they should be taken. Biomesight only cites one B-Vitamin (with no indication of importance). Some ME/CFS studies on the top suggested B-Vitamin ( Vitamin B1, thiamine ) suggested by Microbiome Prescription are shown below.

Your Choice as to Path

IMHO, there is no right answer. Go with Biomesight, Go with what a medical practitioner suggests. Go with whatever you see next in an influencer YouTube.

My best answer is above, it uses a massive amount of data to compute suggestions with a complete evidence trail for people to openly challenge. I have worked professionally as an information auditor and made sure auditability was build into the expert system. I have tuned the expert system to produce good results by doing cross-validation – i.e. 80-90% of suggestions for tested conditions are known to improve that condition from independent clinical studies. In this case, the top suggestions are in agreement with what has been known to help with his specific condition: ME/CFS. MP suggestions are not random shots in the dark but heavily data driven.

It is your choice — just don’t “mix and match” suggestions from different sources.

REMEMBER: Going Biomesight and transferring data to MicrobiomePrescription gives two analysis that you can compare and potentially ask the provider for the basis of their 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 can compute items to take, those computations do not provide information on rotations etc.

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.

Technical Notes: Percentages of Percentiles for Health Measure?

This is a part of a series of Technical Notes on Microbiome Analysis

For a while I have been using a variation of this concept for 16s samples that I have reviewed. The concept is very simple to a statistician:

Percentiles is converting data into a native uniform distribution. If you sample for 1000 boxes where each box has 100 balls numbered 1-100, then you expect the distribution of the balls samples to be uniform. It they are not, then something is definitely unfair.

Concept

With the microbiome things are a little more complex because a high in a single strain may push it species into high and thus the genus into high. We could do independent levels, for example species only or genus only. The problem is that the population size starts to drop and thus the sensitivity decreases as a result.

I happen to have a small collection of shotgun samples processed through CosmosID. Their report give percentile for most of what they measure. Getting accurate percentiles requires large sample sizes.

Below I have charted the results with single percentile ranges from reports that have between 2000 and 5000 different biological units reported. I have charted using different approach (the kitchen sink and then select taxological levels).

All of these samples are from people with health issues. Note that the numbers come from rounding so 100% is just 99.5 to 100 (and not 99.5 to 100.5) so the spikes at 100 is likely twice as high.

Kitchen Sink

Filter to Species Only

Genus Level

Family Level

Bottom Line

Comparing different levels can be informative, to illustrate, we have species below with good uniformity until we hit the high levels.

Looking at the genus level for the same sample, the pattern is very different.

In this case, we drilled down into these high species and got a predominance of Corynebacterium species that fell into our 100% range (99.5-100 percentiles).

Taxonomy NameAbundance
Anaerococcus mediterraneensis0.005611
Anaerococcus prevotii0.006486
Bacteroides rodentium0.001238
Corynebacteriaceae bacterium ‘ARUP UnID 227’0.000437
Corynebacterium ammoniagenes0.000586
Corynebacterium aurimucosum0.1573
Corynebacterium callunae0.00013
Corynebacterium camporealensis0.002243
Corynebacterium casei0.000726
Corynebacterium comes0.000391
Corynebacterium diphtheriae0.0755
Corynebacterium endometrii0.001051
Corynebacterium flavescens0.001684
Corynebacterium humireducens0.00053
Corynebacterium imitans0.001024
Corynebacterium jeikeium0.01813
Corynebacterium lactis0.000437
Corynebacterium liangguodongii0.000558
Corynebacterium minutissimum0.03511
Corynebacterium phocae0.000865
Corynebacterium pseudotuberculosis0.000233
Corynebacterium renale0.000493
Corynebacterium resistens0.001182
Corynebacterium riegelii0.001321
Corynebacterium segmentosum0.007016
Corynebacterium simulans0.3615
Corynebacterium singulare0.01858
Corynebacterium sp. NML 98-01160.001024
Corynebacterium stationis0.000577
Corynebacterium striatum0.04709
Corynebacterium timonense0.001321
Corynebacterium urealyticum0.00107
Corynebacterium uterequi0.000642
Corynebacterium yudongzhengii0.000689
Cutibacterium acnes0.002298
Dehalococcoides mccartyi0.006123
Dermabacter jinjuensis0.01404
Dermabacter vaginalis0.001265
Fastidiosipila sanguinis0.003536
Finegoldia magna0.06368
Helcococcus kunzii0.00014
Homo sapiens1.985
Lawsonella clevelandensis0.003154
Mycobacterium gallinarum0.000261
Mycobacterium sp. DL5920.00013
Mycobacterium sp. ELW10.001107
Mycobacterium sp. EPa450.002298
Mycobacterium sp. PYR150.008328
Mycolicibacterium aichiense0.000223
Negativicoccus massiliensis0.001935
Peptoniphilus harei0.04272
Peptoniphilus sp. ING2-D1G0.000893
Porphyromonas asaccharolytica0.06443
Porphyromonas bennonis0.000521
Propionibacterium freudenreichii0.000465
Schaalia radingae0.001089
Streptococcus pyogenes0.00241
Streptococcus sp. NCTC 115670.000149
Sutterella stercoricanis0.000149
Tessaracoccus timonensis0.00094
uncultured Chroococcidiopsis sp.0.000242
uncultured Rhizobium sp.0.000772

We could also produce single value statistical measures — for example Chi2. We have an a priori expected value of 1% in each bucket.

IMHO, percentages of percentiles is likely more effective in evaluating an individual person’s gut microbiome. It seems to be able to separate the noise from what is significant, for example Corynebacterium cited above where the cause is a proliferation of species and not dominance of one species.

This has since cascaded into an Eubiosis Index.

Microbiome of person with Multiple Sclerosis (after Lyme and EBV)

Back Story

  • In 1983 or 1984 I suffered from EBV (mononucleosis)
  • In 1984 or 1985 – I had appendix removed
  • In 1991 I had a resurgence of fatigue like EBV reactivation, plus apparition of anxiety
  • In 2004, I was bitten by a tick, I thought at the time that it was a spider. Few weeks after the bite, I had flu symptoms who last very long, like months, and some intermittent fever. When I talked about my intermittent fever to doctors, they where looking at me as if I was crazy. Later I learnt that Lyme was in the area.
  • Between 2004 and 2007, lots of weird symptoms appeared. Doctors were saying it was in my head
  • In 2007, I had an urinary tract infection. I took Cipro, and all my little weird symptoms that I had notice for couple of years, have worsened. I started to have mood change, internal tremors.
  • Between 2007 and 2011 -I’ve met 3 neurologists, they said I maybe have multiple sclerosis, even if my MRI at this time were clear.
  • In 2015, another urinary tract infection, Cipro again, symptoms once again worsened.
  • In 2016, I received Multiple sclerosis, (MS), diagnosis.
    • I saw a naturopath. She run urinary test to see organic acid. And she build a protocol. I follow this protocol for 3 months, with no change.
    • I went to see a LLMD in USA for a year with some improvement.
  • Between 2007 and 2011 -I’ve met 3 neurologists, they said I maybe have multiple sclerosis, even if my MRI at this time were clear. They said that I have to wait for another crisis to confirm. But they gave me
  • I now have dysautomia, probably MCAS and SIBO. I also feel sick in transports. I do have intolerance to heat and cold. I have had big constipation problems for years.
    • I started to take Mutaflor[E.Coli Nissle 1917] for constipation. It’s helping.
    • I also started Akkermansia about 1 month ago.
    • B1 (1000mg/day)

We have two test results available: Biomesight and Genova test.

Analysis

The Percentage of Percentiles showed no statistically significant pattern with significance at 0.90 (we look for above .99) to be concerned.

Looking at the Health Analysis,

  • Bacteroides/Clostridium Ratio is very high (97%ile_
  • Anti inflammatory Bacteria Score is high (94%ile)
  • Butyrate is low (1.2%ile)
  • D-Lactic Acid is low [GOOD THING, high levels often are seen with brain fog and cognitive issues)
  • Dopamine, Serotonin are both high (97%ile) – may account for mood issues
  • Hydrogen, Hydrogen sulfide (H2S), Methane are all low with Methane being the highest (46%ile), so traditional SIBO is unlikely.

Potential Medical Conditions Detected

The following were flagged in agreement with her history:

  • ME/CFS without IBS
  • Fibromyalgia
  • Mood Disorders
  • COVID-19

And last, Intelligence at 91%ile which agrees with details from emails.

And for Bacteria deemed Unhealthy we have quite a few.

NameRankPercentileCountCommentMore Info
[Ruminococcus] gnavusspecies9854710Not Healthy PredictorCitation
Actinomycesgenus90250PathogenCitation
Anaerotruncus colihominisspecies863070Not Healthy PredictorCitation
Bacillusgenus92170PathogenCitation
Blautia productaspecies9712950Not Healthy PredictorCitation
Collinsellagenus00High COVID RiskCitation
Doreagenus9937010Increased COVID riskCitation
Eggerthella lentaspecies951470Not Healthy PredictorCitation
Lactobacillusgenus89960PathogenCitation
Legionellagenus89130include notable pathogensCitation
Ligilactobacillus salivariusspecies87420Not Healthy PredictorCitation
Staphylococcus aureusspecies7540Skin infections, sinusitis, food poisoningCitation
Staphylococcus haemolyticusspecies89130PathogenCitation
Streptococcus australisspecies82210Not Healthy PredictorCitation
Streptococcus oralisspecies6640Infectious bacteriaCitation
Streptococcus sanguinisspecies8780Not Healthy PredictorCitation
Streptococcus vestibularisspecies67380Infectious bacteriaCitation
Veillonella atypicaspecies76170Not Healthy PredictorCitation

I looked at her GI Effects test with the new Conditions matching (See this post) and nothing was identified by pattern matching.

Using Jason’s criteria, we see that there is a long way from health improvement.

Going Forward

With a diagnosis of Multiple sclerosis, I was curious to see the degree of pattern matching to published studies. She is at the 88%ile (i.e. her pattern the reported pattern better then 9 out of 10 people).

While it appears that Lyme played a role, the literature is very sparse on Lyme and she has no matches

Strategy

I will do the usual “Just give me suggestions’ (4 ways of picking bacteria) and then add in:

  • Multiple Sclerosis
  • Mood Disorders

This gives us 6 algorithms to build suggestions from. To which we add the new one to hand pick and then process. So we have 7 algorithms being used.

Review of Suggestions

My first curiosity is where does Cipro (Ciprofloxacin) set in suggestions. It is at a positive 275 our of 494. The top antibiotic is amoxicillin which is used for both ME/CFS and Lyme disease.

I was curious if there is a MS connection to either of these antibiotics and found Antibiotic Use and Risk of Multiple Sclerosis [2006] which contains a variety of gems:

  • “use of penicillins(includes amoxicillin) in the 3 years before the index date decreased the risk of developing a first attack of multiple sclerosis (odds ratio = 0.5, 95% confidence interval: 0.3, 0.9 for those who used penicillins for ≥15 days compared with no use).”

For Cipro, I found no equivalent studies and some social media claiming that Cipro triggered MS in themselves.

No probiotic made it above the threshold except a particular mixture: bifidobacterium pseudocatenulatum li09,bifidobacterium catenulatum li10 (probiotics). I currently know of no retail source for this mixture (but can see a lot of studies). Neither can I locate any retail products with any form of bifidobacterium pseudocatenulatum or bifidobacterium catenulatum.

Questions and Answers

Q: In my history, you don’t seem to take into account the positive tests for borrelia and babesia, but only the diagnosis of multiple sclerosis. Am I mistaken? And is it because there are few studies on Lyme disease in relation to the microbiota? 

Q: There is mention of human milk but nothing about dairy. I’m wondering if goat cheese is ok. I consume goat cheese from time to time and wondered if it’s good or bad.

  • Human milk contains different sugars than goat or sheep or cow or camel or… I have data on goat and cow. Most studies have been done on using them for yogurts which alters their composition.
  • Looking at the details (see YouTube video), all dairy are negative (not greatly often, but consistently negative for different dairy products), so reduce or eliminate.

Q: Does acacia fiber is consider oligofructose-enriched inulin ? I’m a bit lost. I bought acacia and wonder if it’s ok.

  • Acacia fiber (a.k.a. gum) is different. There is a study comparing them, PREBIOTIC EFFECTS OF INULIN AND ACACIA GUM [2015]. Acacia fiber was not in the list for to take or to avoid, so no known harm nor benefit (apart from the usual impact on the pocket book)

Q: In the recommendations, it’s said to avoid whole grain wheat. But does it include einkorn and buckwheat ?

  • No, buckwheat is not wheat, it is a seed (just like peanut is not a nut) — English can be very misleading at time!!! While it is true that Einkorn is the most primitive form of wheat on Earth, modern wheat (which is what the clinical studies used) is sufficiently different in content. “Einkorn kernels have higher protein, antioxidant (carotenoids and tocols), fructans and monounsaturated fatty acids content” [2013]. Many of those changes will cause a different effect on the microbiome.
    These are slight negative (see video), I would not be concerned about this.

Q: Alan McDonnald’s work shows that all the patients he tested with a diagnosis of multiple sclerosis were positive for at least one strain of borrelia, in addition to having their EBV reactivated. This is generally the case with Lyme. And since I’m treating Lyme, I have a lot of symptoms who alleviate.

  • Unfortunately, 16s tests do badly with detecting that bacteria. Shotgun tests are 10 to 40x better at detecting this bacteria. Some level may be present in 30% if the population. See this page

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 can compute items to take, those computations do not provide information on rotations etc.

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.

Technical Note: With the same set of samples from the same labs you can get very different averages!

This post originated from a dialog with a Ph.D. in Molecular Genetics that I often discuss many aspects of microbiome analysis with.

The root of the problem is how many “Reads” from a 16s sample do you deem to be threshold for reliability. A “Read”, “num_hits” or “Count” is the number of matches to specific pattern found in the sample that matches a library. These are “best efforts” identification. Not always correct.

Accuracy can be as low as 62% [Then and now: use of 16S rDNA gene sequencing for bacterial identification and discovery of novel bacteria in clinical microbiology laboratories ]. It is generally assumed that a single “Read” is questionable. Commercial labs and test providers will often use them so they can claim that they identify more bacteria than the competition. Accuracy is rarely a marketing concept.

To this end, we processed the biggest collection of samples of one lab with different Read Levels to see what happens. The higher accuracy required to be included that you use, the higher the values.

obsmeanstddevmedianboxplotlowboxplothightax_namerankReads
471386.55301.7301070Neisseriagenus1
242733.67387.05010110Neisseriagenus2
1361275.89835.57030210Neisseriagenus3
951800.311747.68020300Neisseriagenus4
682491.213853.41200360Neisseriagenus5
553059.515375.21600380Neisseriagenus6
414071.717748.52000500Neisseriagenus7
305517.020650.42400778Neisseriagenus8
217825.724488.4350501532Neisseriagenus9

Two labs may report different reference ranges for the simple reason that one requires at least 2 reads and the other lab 4 reads. This decision is often well hidden from the consumer. If the reference ranges are based on 4 reads and you apply them to 1 read samples then you will get a lot of false too high and too lows.

For the above example bacteria a 1 read reference range would have 386 being the average, while a 4 read reference range would have the average being 1800. So, a sample with 800 from 2 reads would be 2x the average for one reference range and and 1/2 the average for the other reference range.

This is part of the complexity of doing microbiome analysis and understanding the mechanism involved. Mechanisms that are often not understood by the labs and kit providers.

Son and Daughter with Autism Analysis – Part 2

This is a follow up the earlier blog post: Son and Daughter with Autism Analysis from a year ago. There has been a lot of changes of the site and revisions of algorithms.

Comparing Siblings

We know from studies that members in the same family often share about 27% of the same strains. Unfortunately with 16s tests (Biomesight, Ombre), we do not get strain information just species information.

Using the new refactor citizen science symptoms (see New Special Studies on Symptoms ), we are presently surprised! We have many forecast symptoms being the same which supports the observation cited above of share taxa, likely at the strain level.

It does hint that less time with each other and a lot more time with other (ideally normal) children may have benefits to the microbiome. Some of the changes may be connected to gender:

  • About twice as many women as men experience depression [Mayo]
  • Increased inflammation is seen in the periphery in both depression and fatigue. [2019] which agrees with the daughter having a lower Anti inflammatory Bacteria Score
MeasureDaughterSon
Anti inflammatory Bacteria Score25.6%ile89%ile
Buytrate Bacteria Score95.9%ile78%ile
Histamine Producers21.8%ile15.3%
Autism From PubMed53/97 (1%ile)
Prior: 54/97 (1%ile)
73/97 (11%ile)
Prior: 53/97 (1%ile)
Forecast SymptomsOfficial Diagnosis: Depression
28 % match on 7 taxa

DePaul University Fatigue Questionnaire : Blurred Vision
25 % match on 8 taxa
Neurological-Sleep: Inability for deep (delta) sleep
23 % match on 13 taxa

Age: 10-20
17 % match on 23 taxa

DePaul University Fatigue Questionnaire : Forgetting what you are trying to say
16 % match on 31 taxa

Next looking at Percentages of Percentiles, we see significant differences. Unfortunately, we do not have gender and age reference tables, so interpretation is fuzzy.

Potential Medical Conditions Detected had nothing significant for either child. Both are at 95.6%ile on Dr. Jason Hawrelak Recommendations (they were 98.9 and 99.7%iles before) .

Detail Comparison

The thing that stands out is that the Son has a lot more Enzymes out of range (with the resulting substrates(consumers) and products also being out of range).

CriteriaDaughterSon
Lab Read Quality6.710.9
PubMed Bacteria Matches for Autism1%ile (53/97)11%ile (73/97)
Outside Range from JasonH55
Outside Range from Medivere1818
Outside Range from Metagenomics99
Outside Range from MyBioma99
Outside Range from Nirvana/CosmosId2323
Outside Range from XenoGene5151
Outside Lab Range (+/- 1.96SD)1122
Outside Box-Plot-Whiskers9876
Outside Kaltoft-Moldrup132248
Bacteria Reported By Lab842757
Bacteria Over 99%ile1311
Bacteria Over 95%ile2845
Bacteria Over 90%ile6661
Bacteria Under 10%ile66285
Bacteria Under 5%ile22214
Shannon Diversity Index3.0642.807
Simpson Diversity Index0.070.088
Chao1 Index2979124924
Rarely Seen 5%6198
Pathogens3635
Kegg Compounds Low9731001
Kegg Compounds High43162
Kegg Enzymes Low89265
Kegg Enzymes High98381
Kegg Products Low55152
Kegg Products High52209
Kegg Substrates Low46148
Kegg Substrates High58229

Looking at KEGG Derived Probiotic suggestions, the list is full of the soil based bacteria found in Prescript-Assist®/SBO Probiotic or Energybalance / ColoBiotica 28 Colon Support or General Biotics/Equilibrium. There was no probiotic above my usual threshold from the consensus, so the above seems to be worth a try.

KEGG Suggested supplements has nothing significant for the daughter, but for the son we have the following being very significant:

  • Serine
  • Threonine
  • Glutamine
  • Cysteine
  • Arginine

A complex amino-acid supplement may be worth an experiment.

As an experiment (and trying to avoid two different kid diet), I did an uber-consensus from each child’s with tons of prescription medication but only one thing above my usual 50% of highest value.

Son Compared to Prior Sample

We can see the spike in low percentile bacteria. This raises the question, has he had COVID (or a COVID vaccine) prior to the sample being done. These spikes show themselves also via Kaltoft-Moldrup and Box-Plot-Whiskers which are both sensitive to this pattern.

CriteriaCurrent SampleOld Sample
Lab Read Quality10.94.4
Outside Range from JasonH44
Outside Range from Medivere1717
Outside Range from Metagenomics1010
Outside Range from MyBioma1313
Outside Range from Nirvana/CosmosId2727
Outside Range from XenoGene4949
Outside Lab Range (+/- 1.96SD)2222
Outside Box-Plot-Whiskers76100
Outside Kaltoft-Moldrup24889
Bacteria Reported By Lab757708
Bacteria Over 90%ile6182
Bacteria Under 10%ile28526
Shannon Diversity Index2.8072.451
Simpson Diversity Index0.0880.15
Chao1 Index2492419183
Lab: Thryve
Pathogens3530
Condition Est. Over 90%ile20
Kegg Compounds Low10011048
Kegg Compounds High162132
Kegg Enzymes Low265115
Kegg Enzymes High381296
Kegg Products Low15274
Kegg Products High209191
Kegg Substrates Low14869
Kegg Substrates High229212
Anti inflammatory Bacteria Score89.2%ile83.2%ile
Buytrate Bacteria Score77.9%ile90.2%ile
Histamine Producers15.3%ile38.2%ile
Histamine dropping is usually a good sign

From this weekend update of special studies, we can get a count of bacteria shifts strongly associated to symptoms.

  • Old Sample: 32 taxa
  • Latest Sample: 60 taxa

Daughter Compared to Prior Sample

First the numbers which are usually similar to the prior sample.

CriteriaCurrent SampleOld Sample
Lab Read Quality6.73.1
Outside Range from JasonH66
Outside Range from Medivere1919
Outside Range from Metagenomics77
Outside Range from MyBioma1212
Outside Range from Nirvana/CosmosId2626
Outside Range from XenoGene4747
Outside Lab Range (+/- 1.96SD)1161
Outside Box-Plot-Whiskers98203
Outside Kaltoft-Moldrup132134
Bacteria Reported By Lab842852
Bacteria Over 90%ile66202
Bacteria Under 10%ile6610
Shannon Diversity Index3.0643.411
Simpson Diversity Index0.070.028
Chao1 Index2979135210
Lab: Thryve
Pathogens3635
Condition Est. Over 90%ile00
Kegg Compounds Low9731027
Kegg Compounds High4380
Kegg Enzymes Low8944
Kegg Enzymes High98171
Kegg Products Low5529
Kegg Products High5286
Kegg Substrates Low4626
Kegg Substrates High58111
Anti inflammatory Bacteria Score25.5%ile28%ile
Buytrate Bacteria Score95.9%ile74.5%ile
Histamine Producers21.7%ile28.7%ile

From this weekend update of special studies, we can get a count of bacteria shifts strongly associated to symptoms.

  • Old Sample 53
  • Latest Sample: 39

Out of curiosity, I compared the symptom associated outliers. We found 3 are matches (of these 39) and one not matches for the taxa reported for each. That is close to the expected percentage of the same strains for people in the same house.

Bacteria NameDaughterSon
  NegativicutesToo HighToo High
  AcidobacteriiaToo LowToo Low
  Bacteroides eggerthiiToo HighToo Low
  PropionibacterialesToo LowToo Low

Going Forward

Autism has challenges because of its complex nature. This is compounded by a low number of samples to work from for Citizen Science analysis. The shifts reported from PubMed have a high pattern match with people who do not have autism.

I am going to try building a consensus for each by doing two itemsL

  • “Just give Me Suggestions”
  • Doing PubMed Autism on [Changing Microbiome]
  • [All Bacteria identified by special studies]

The rationale is that the last one identify the bacteria that appears to be symptom causing in many people. We have a very poor match from what we do have a match for. This is not surprising because autism is a very wide spectrum.

We then see six sets of suggestion

Son

When I look at the details we have over 150 items with 6 recommended take (i.e. everyone agrees)

The probiotics that have no known adverse risk for any bacteria is below. The high value is 510.

Daughter

When I look at the details we have just 15 items with 6 recommended take (i.e. everyone agrees)

The probiotics that have no known adverse risk for any bacteria are low in computed benefit, so I would ignore them.

  • Their values are low: 16/31 out of a high value of 301

Bottom Line

The failure to find significant matching patterns is a bit of a frustration to me. What we did find had very good agree for the son with 150+ items having each of the size suggestion set agreeing for the take. For the daughter, it was not as strong: 15 for 6 sets being in agreement, and 50 with 5 sets being in agreement.

Questions

  1. I assume higher anti-inflammatory score is better – Daughter was 25% and Son was 89%
  2. Deep Sleep with Son – 23% match that he has deep sleep issues is pretty strong?
    • Does not jump out, but indicates that microbiome is playing a role.
  3. Son – lot more missed enzymes – what is that do you believe and probiotics help with that?
    • I avoid the word “believe”. A rational assumption is that disruption of enzymes compare to others impacts how the cells (including brain cells) react.
  4. Spike in low percentage bacteria – likely long Covid for Son means he has less good bacteria now?
    • I avoid the words “good” or “bad” bacteria. Any bacteria far enough from typical values become bad; disruption to the microbiome and the body. Theses spikes are typically seen (pattern matching) with two conditions: Long COVID and ME/CFS. A common symptom of these two issues are cognitive issues – for example: memory, ability to learn, etc.
  5. Histamine – Higher percentage is worse correct?  Daughter was 21% and son was 15% 
  6. Higher Butryate percentage is better?  95% daughter / 78% son

The Journey Begins with your microbiome

Thanks for joining me!

This is a companion site to the analysis site at: 

https://microbiomeprescription.com/

The intent of this site to assist people with health issues that are, or could be, microbiome connected. There are MANY conditions known to have the severity being a function of the microbiome dysfunction, including Autism, Alzheimer’s, Anxiety and Depression. See this list of studies from the US National Library of Medicine. Individual symptoms like brain fog, anxiety and depression have strong statistical association to the microbiome. A few of them are listed here.

The base rule of the site is to avoid speculation, keep to facts from published studies and to facts from statistical analysis(with the source data available for those wish to replicate the results). Internet hearsay is avoid like the plague it is.

Open data and Open source are our mottos!

Continue reading “The Journey Begins with your microbiome” →Posted on  by lassesenEdit

Biomesight #4 Sample: IBS and COVID

We have a varied history with some storms blowing us off courses. Here’s a list of the tests and prior blog posts:

His comments are short:

  • I would say some small subjective improvements since last time, but no major changes. Reminder: I have a friendly MD in terms of antibiotics.
  • Metronidazole was on top in the last samples, I did it back then.
    • Comment: Metronidazole is no longer at the top but dropped down to 16% of the highest value. It appears to have done its magic in reducing the bacteria pointing to it as a tool..

Base Analysis

When people have multiple samples, I like to do side-by-side comparisons, especially when someone has been doing some of the suggestions suggested. The suggestions are computed and may not always work. Expert Systems and AI are not perfect; they typically do better than a person with only a few years of experience that has training in the discipline (better consistency, remember more facts, etc). How are we doing objectively?

Scores

We see two positive shifts in the latest sample: Increase of Anti inflammatory Bacteria Score and decrease of Histamine Producers.

Percentages of Percentiles

We see a lot of bouncing around between samples. The middle two images matches the typical pattern seen with ME/CFS and Long COVID. Those shifts have faded over the last 3 months with a different pattern appearing indicating a different dialect of gut dysfunction.

Multi-Vector Comparison

The main numbers are below. The take away, less bacteria that are in the high percentile range (at 95%ile, 10 -> 28 -> 23 -> 8). The numbers bounce around with the middle two being similar and the other two also similar. There are no really clear shift in these measures.

Criteria11/18/20215/20/20226/22/20239/4/2023
Lab Read Quality8.15.54.77.2
Outside Range from JasonH6699
Outside Range from Medivere16161515
Outside Range from Metagenomics8877
Outside Range from MyBioma5566
Outside Range from Nirvana/CosmosId20202323
Outside Range from XenoGene29293535
Outside Lab Range (+/- 1.96SD)76173
Outside Box-Plot-Whiskers36695438
Outside Kaltoft-Moldrup93484788
Bacteria Reported By Lab652508542558
Bacteria Over 99%ile7462
Bacteria Over 95%ile1028238
Bacteria Over 90%ile29423622
Bacteria Under 10%ile2084150175
Bacteria Under 5%ile180198157
Shannon Diversity Index1.8531.8261.2721.556
Simpson Diversity Index0.0560.0380.0870.09
Rarely Seen 1%2271
Rarely Seen 5%145218
Pathogens41242936

From Special Studies

The top match was the same on all of the samples, with an increase when there was actually COVID.

Criteria11/18/20215/20/20226/22/20239/4/2023
COVID19 (Long Hauler)28%ile33%ile41%ile28%ile
Next one:15%ile26%ile20%ile13%ile

The “next one” dropping implies some reduction of dysbiosis

Health Analysis

Using Dr. Jason Hawrelak Recommendations, there are many items on the edge of being in range with some items of interest (I strike out those that are unlikely to be of great concern):

  • Faecalibacterium prausnitzii at 27% of the microbiome or 96%ile
  • Akkermansia — 0.009 % of the microbiome or 35%ile
  • Bifidobacterium 0.016 % of the microbiome or 16%ile
  • Bacteroides – 27% of microbiome, or 64%ile

Additionally, two indicate increased risk of Candida (new feature just added)

  • Phocaeicola dorei at 10% of the microbiome or 91%ile
  • Faecalibacterium prausnitzii at 27% of the microbiome or 96%ile

I would suggest a test for candida to be safe. The data suggests a risk. If confirmed, candida would contribute significantly to gut dysbiosis [The interplay between gut bacteria and the yeast Candida albicans[2021]). I did a “back-flip” check of the top prescription items, and all of them reduces Candida (studies cited below).

Addendum – Predicted Symptoms

This was just added to the site today as a further refactor based on New Special Studies on Symptoms data. These are from [My Profile Tab]

Criteria11/18/20215/20/20226/22/20239/4/2023
Forecast Major SymptomsNeurological: Cognitive/Sensory Overload
40 % match on 25 taxa

DePaul University Fatigue Questionnaire : Racing heart
38 % match on 13 taxa

DePaul University Fatigue Questionnaire : Difficulty falling asleep
37 % match on 27 taxa

DePaul University Fatigue Questionnaire : Difficulty finding the right word
35 % match on 20 taxa
Autonomic Manifestations: urinary frequency dysfunction
66 % match on 6 taxa
Immune Manifestations: Bloating
37 % match on 45 taxa

Neurological-Audio: hypersensitivity to noise
35 % match on 28 taxa
NoneNeurological-Sleep: Chaotic diurnal sleep rhythms (Erratic Sleep)
50 % match on 18 taxa

Neurological: Spatial instability and disorientation
37 % match on 16 taxa

This can be helpful for judging possible severity (and potential improvement of some symptoms), for example: Neurological: Cognitive/Sensory Overload. See [Special Studies] tab.

  • 2021 – 40% matches
  • 2022- 24% matches
  • 6/22/23 – 16% matches
  • 9/4/2023 – 4% matches

Going Forward

COVID has had quite an impact on this microbiome. I am going to just go with the “Just Give Me Suggestions” option with the addition of what matched his diagnosis:

  • Irritable Bowel Syndrome  (68 %ile) 7 of 68

To explain a bit more. First I click the button below

And then click I could click the consensus report to see what the top items are:

Which are shown below.

In this case, I want to add Irritable Bowel Syndrome suggestions (on the Changing Microbiome Tab)

Instead of the usual 4 packages of suggestions, we have 5

When we look at the consensus report we see the same items there, but the values have increased.

The intent is put a little bias on the numbers towards specific conditions of greatest concern.

PDF Suggestions

I tend to favor the PDF suggestions because it simplifies things for many readers. Also the PDF gives a good list of citations (never complete) used to make the citations to persuade MDs to see that the suggestions are based on studies — a lot of studies.

The PDF suggestions are below (using the consensus view is another option for those more technically orientated). I clip from the PDF to keep the blog simpler for the typical reader.

This is a little longer list than usual, so I went to the consensus report to get priority data. Top value was 618, so 309 is the 50% threshold.

These appear to be of low influence with the exception of l.bulgaricus:

Minor note: quercetin with resveratrol is an avoid, quercetin is a take. resveratrol by itself is a negative (-113). At times, you need to look at the technical details/consensus to clarify things; the data we are using is incomplete and sparse…. If clearly contradictory suggestions appear, then don’t do them (thing an abundance of caution).

Because he has an antibiotic friendly MD, the following are the TOP antibiotics with notes:

CFS Antibiotics are also above the threshold. Since the prior sample had a strong Long COVID or ME/CFS Profile, I would be inclined to include one of those below in the antibiotic rotation. The microbiome cannot make a diagnosis of most things, with most ME/CFS microbiomes there is a particular pattern which you had in your last sample but which has disappeared from your current sample which looks more like your first sample. I read this as recovering from ME/CFS….  in likely a fragile state since relapse is very common with ME/CFS.

My own experience is that it is better to overcure ME/CFS and when there are signs of recovery…. no backflips of joy or running marathons; keep doing slow walks that becomes a bit further each week for 6-12 months. Your microbiome is fragile and can quickly slip back.

I prefer to use the strategy of going for prescription items that are both suggested from the microbiome and been shown to help with one or more of the diagnosis conditions. This usually encounter low resistance from physicians — they are clueless for the microbiome, but very accepting of published studies. An antibiotic that is used as a prophylaxis usually encounter little resistance.

KEGG Suggestions

This is done by using information from the bacteria found with some fudge factors. I am in discussion with some Ph.D. candidates to build this concept directly from the FASTQ files and will hopefully have this as an added feature next year.

The KEGG probiotics is the usual pattern for ME/CFS and Long COVID with the top one being the usual, with the top reasonably available ones for other families shown below. I usually like to compare the values with those from consensus to minimum risk (i.e. two thumbs up, we do; mixed, we skipped)

KEGG Supplements

From the list, we will look only at those with a z-score (statistical significance) over 2. After each we put the consensus value (if it is listed)

Only two items are with high confidence.

How to Proceed Suggestions

The suggestions should be thought as influencers. The human population is often a good analogy or parable for the microbiome population. Each influencer shifts the population in the desired direction. Based on Cecile Jadin’s work and several studies, I am a firm believer in short duration (1-2 weeks) of each influencers. Just as with human influencers, people stop listening if the same person just keeps droning on and on. If a different person starts speaking, you get persuaded more. If a mob start to shout, yet a different human behavior will occur. In terms of the microbiome, “stop listening” means mutations that are resistant to the item will start to increase. Items line vitamins and minerals can be taken continuously; items that are likely to have bacteria resistance developed should be taken for a week and then another item replace it.

The items to rotate:

  • Antibiotics listed above
  • Probiotics: lactobacillus salivarius and lactobacillus bulgaricus
  • Herbs and spices: cinnamon, ginger, black cumin, thyme, rosemary, quercetin (suggests just before each antibiotic with a few days of overlap because it has potential synergistic activity with antibiotics [2020], [2016],[2018] )

Remember our goal is to destabilize a stable microbiome dysfunction.

Questions and Answers

While there has not been significant changes in many of the vectors between this sample and the prior sample from a few months earlier, there has been two significant objective changes:

  • Significant improvement of Anti inflammatory Bacteria Score (higher) and Histamine Producers (Lower).
  • The lost of the ME/CFS – Long COVID spike in the 0-9%ile

Q: Do you/should I use the colored list now instead of the consensus list?

  • Either are fine, the color list (from PDF) is what I tend to use in post because it is easier for new readers to understand (and automatically sent on new uploads). The consensus page is more complex but allows people to apply their own logic and priorities.

Q: “Quercetin (suggests just before each antibiotic with a few days of overlap because it has potential synergistic activity with antibiotics”

Q: I just did Mutaflor for 8 days and felt really tired all the time (but in the end I also got a flu/cold, so maybe that was the reason and not mutaflor). Nevertheless, if it was a herx reaction, I wonder if I should have taken it for longer until the reaction disappeared? (I stopped it 4 days ago.) Not sure if this question even makes sense.

  • My personal choice would be to keep taking it for at least a week (perhaps 2). Remember that the traditional pattern for a herx is feeling bad for X hours and then things get better. The duration of the feeling bad usually decrease from day to day. Catching a cold makes interpretation challenging.

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 can compute items to take, those computations do not provide information on rotations etc.

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.

Technical Note: Prevalence, Average and Not Reported

In reviewing many papers with the microbiome I noticed that often the researchers restrict their examinations to the taxa that is reported in all samples. I suspect this is due to a lack of sufficient statistical training and/or not understanding the natures of the microbiome.

Recently I came across these papers that uses an approach that I often have used, working off relative frequency of detection a.k.a. prevalence.

This post is going to use samples available at Microbiome Prescription Citizen Science site. We are going to restrict to one lab source and divide the data into two groups based on their self-declare symptoms and diagnosis.

  • Patients with Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) declared [Obs: 271]
  • Patients without Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) declared and other status declared (for example: “Asymptomatic” [Obs:569]

Naive First Pass

We are going to take the average count for each group ignoring no values reported. We are going to restrict it to taxa where we have at least 30 non-zero values [1,564 taxa]. We found some 77 taxa with a t-score over 2.81 (p < 0.005)

taxa nametaxa rankShiftT_score
Prevotella coprispecieslow in ME/CFS-5.27
Prevotellagenuslow in ME/CFS-4.52
Sporolactobacillaceaefamilylow in ME/CFS-4.2
Sporolactobacillus putidusspecieslow in ME/CFS-4.19
Sporolactobacillusgenuslow in ME/CFS-4.19
Prevotellaceaefamilylow in ME/CFS-4.1
Firmicutesphylumhigh in ME/CFS3.94
Blautiagenushigh in ME/CFS3.91
Cetobacterium cetispecieshigh in ME/CFS3.89
Cetobacteriumgenushigh in ME/CFS3.84

Deeming Not Reported to be Zero

In this case we have 78 taxa with a t-score over 2.81 with slight changes of t-scores.

taxa nametaxa rankShiftT_score
Prevotella coprispecieslow in ME/CFS-5.31
Sporolactobacillaceaefamilylow in ME/CFS-4.63
Sporolactobacillus putidusspecieslow in ME/CFS-4.62
Sporolactobacillusgenuslow in ME/CFS-4.62
Prevotellagenuslow in ME/CFS-4.5
Prevotella oulorumspecieslow in ME/CFS-4.35
Prevotellaceaefamilylow in ME/CFS-4.08
Bifidobacterium gallicumspecieslow in ME/CFS-3.97
Firmicutesphylumhigh in ME/CFS3.94
Blautiagenushigh in ME/CFS3.91

Prevalence

We followed the same process as above and limited things to a Chi-2 probability of < 0.005 (as used above) We ended up with 65 taxa.

tax_NameTax_RankPrevalence
in MECFS %
Prevalence
Control %
DifferenceChi2FoldChange
Deferribacteresphylum33.62013.5141.7
Erysipelothrix inopinataspecies2110.710.3142
Deferribacteralesorder33.62013.5141.7
Deferribacteraceaefamily33.62013.5141.7
Deferribacteresclass33.62013.5141.7
Mogibacterium vescumspecies27.715.811.9131.8
Haploplasma cavigenitaliumspecies8.52.85.7133
Haploplasmagenus8.52.85.7133
Gluconobactergenus15.16.98.3132.2
Prosthecobacter fluviatilisspecies7.72.55.3123.1

Comparing these two lists, we found only 6 taxa in common

  • Bifidobacterium angulatum
  • Propionigenium modestum
  • Pseudomonas viridiflava
  • Cetobacterium ceti
  • Cetobacterium
  • Propionigenium

The next result is that we have 78+65 – 6 = 137 statistically significant bacteria with p < 0.005.

Bottom Line

There are at least two different statistical ways of determining significance. IMHO, the prevalence approach is likely to be a superior tool for diagnosis purposes because it is possible to compute the probability of a match to the above patterns despite some bacteria not being reported.

The full list of bacteria is listed here.

Technical Notes: Statistics and Diversity Indices and PofP

A reader raised a valid question which actually triggers other related questions.

You seem to like the “percentage of percentiles” measurement, but I’m not convinced it’s being analyzed appropriately. As I understand it, you first convert to percentiles, getting numbers in [0, 100]. I think this is fine. Then you histogram these percentiles. Because each lab will perform the same measurements every time, I think this is also fine. However, the result is compositional data in the sense of Aitchison, and it should be analyzed in a manner consistent with that. For compositional data, a chi^2 test is inappropriate because it relies on the number of species (or genera) measured.

My suggestion is to apply a centered logratio transform to each person’s percentages and fit a normal distribution to the transformed data. To determine whether someone’s microbiome deviates significantly, calculate a multivariate normal tail probability. Beware that the covariance matrix will be rank deficient (you’re in a ten-dimensional space, but there are only nine parameters because percentages sum to 100). You may want a robust fit because it’s reasonable to expect that the microbiome of someone ill might be an outlier.

For more information about compositional data, see Aitchison, J., “The Statistical Analysis of Compositional Data,” Journal of the Royal Statistical Society. Series B (Methodological) Vol. 44, No. 2 (1982), pp. 139-177; Aitchison, J., “The Statistical Analysis of Compositional Data,” Chapman & Hall, London, 1986; and Aitchison, J. “A Concise Guide to Compositional Data Analysis,” unpublished manuscript, 2005, available online (just Google). For other approaches to compositional data analysis, see Greenacre, Michael; Grunsky, Eric; Bacon-Shone, John; Erb, Ionas; Quinn, Thomas, “Aitchison’s Compositional Data Analysis 40 Years On: A Reappraisal,” arXiv:2201.05197, 13 Jan 2022, to appear in Statistical Science.

What is the statistical basis for other Diversity Indices?

How to calculate these numbers is well determined — they seem to be brilliant ideas tossed out there that seems to fit the data for some study. For some background, see this page. The problem is a lack of rigor, especially statistical rigor.

Diversity indices, particularly the Shannon-Wiener index, have extensively been used in analyzing patterns of diversity at different geographic and ecological scales. These indices have serious conceptual and statistical problems which make comparisons of species richness or species abundances across communities nearly impossible. 

Conceptual and statistical problems associated with the use of diversity indices in ecology [2009]

The problem is an absence of a native statistical model. For example, it does not fit the usual ones.

The key question is simple, what is the distribution underlying diversity Indices? We read ” In the literature of biodiversity, according to Ricotta (2005), there are a “jungle” of biological measures of diversity.”[2017]. Zheng’s A new diversity estimator[2017] in Journal of Statistical Distributions and Applications where he states “There are many other open problems built on this connection between birthday problem and diversity measures. ” The problem is this, the birthday problems deals with 366 discrete well defined boxes that are well defined. Dealing with the microbiome, we lack these boxes. Consider a measure of a microbiome sample in 2000, there are a large number of different bacteria species in Lactobacillus. Today, we have these species no longer placed in 1 genus, but 25 genus [2020] including:

  • Acetilactobacillus,
  • Agrilactobacillus,
  • Amylolactobacillus,
  • Apilactobacillus,
  • Bombilactobacillus,
  • Companilactobacillus,
  • Dellaglioa,
  • Fructilactobacillus,
  • Furfurilactobacillus,
  • Holzapfelia,
  • Lacticaseibacillus,
  • Lactiplantibacillus,
  • Lapidilactobacillus,
  • Latilactobacillus,
  • Lentilactobacillus,
  • Levilactobacillus,
  • Ligilactobacillus,
  • Limosilactobacillus,
  • Liquorilactobacillus,
  • Loigolactobacilus,
  • Paucilactobacillus,
  • Schleiferilactobacillus,  
  • Secundilactobacillus.

With the same strains/species, our diversity indices will be very different because our boxes are arbitrary and “soft” unlike the days of the year or the roll of a dice.

Back to percentage of percentiles

While I show genus and species in the table for ease of understanding of the typical reader, I originally did it solely with the lowest identifiable levels (the “atoms” or the microbiome) – species. At the species level, it is not compositional. There is no composition! Looking at the data that was actually received, I noticed many genus had no species listed. In some cases, the genus had species, but none of the known ones were detected. In other cases, the test did not report any species in over 3000 test results.

On this basis I decided to use try using both species and genus. I soon discovered that they almost always exhibit a similar pattern and chi^2. At this point, I opted for benefiting my readers and not as much rigor as some would like. We could do the lowest taxonomical level reporting across the hierarchy as one solution.

This approach ends up with us side-stepping the classification issues cited above. We are dealing with distinctive, non-overlapping events (a bacteria being identified) and then convert them to percentile giving use a continuous uniform distribution for each of these independent events. IMHO, at this point we have a good model to chi^2 test. We are not dealing with measuring a population, just a sample.

In answer to “a chi^2 test is inappropriate because it relies on the number of species (or genera) measured.” is missing the point. If I get two bags of coins from the bank and then flip them to determine if they are biased — whether the bags contains 1000 or 100,000 coins is significant only on the ability to determine the margin of error. The number of species/genus is only significant in that sense. If there is a strong bias with a small number, then having more will not change the bias.

Technical Note: Lab Quality Versus Bacteria Reported

For samples coming from uBiome, Ombre/Thryve and Biomesight there are two important numbers reported. [Count] and [Count_Norm].

  • [Count] is the number coming from the lab equipment, the bacteria detected.
  • [Count_Norm] is the above number scaled to be out of one million (1,000,000)

The question arises, if you have low lab quality and the number of bad bacteria also dropped: Is this an actual improvement or a false improvement due to low lab quality?

To partially answer that question, I pulled biomesight samples (biggest collection) and plotted the data. Shown below:

  • Vertical axis is number of bacteria reported
  • Horizontal axis is lab quality measure

As is shown, there is a relationship.

Using this data and sample values of 4.3 and 8.4, we compute 546 and 643 for expected bacteria (just use the formula on the chart). This means that with 4.3 we expect only 85% (546/643 * 100) of the count seen with 8.4.

So we see the changes below are likely true improvements: (Left is 4.3, right is 8.4)

For bacteria reported by lab, the numbers suggests the left column has less odd bacteria and the gut microbiome may becoming more uniform.

For anyone interested in doing their own charting and analysis, the raw data is https://citizenscience.microbiomeprescription.com/