lassesen 11:52 am on July 4, 2018  

End Product Predictions vs New CFS Study

A reader sent me a link to a new study, July 3rd, 2018 “Insights into myalgic encephalomyelitis/chronic fatigue syndrome phenotypes through comprehensive metabolomics“. Authors include some well known CFS researchers: Nagy-Szakal D1, Barupal DK2, Lee B1, Che X1, Williams BL1, Kahn EJR1, Ukaigwe JE1, Bateman L3, Klimas NG4,5, Komaroff AL6, Levine S7, Montoya JG8, Peterson DL9, Levin B10, Hornig M1, Fiehn O11, Lipkin WI12.

Full text is here. WARNING: Heavy data science language used.

In the supplemental material was a table of compound names and shifts in CFS patients:

Study2

I was curious to see how that compared to predictions from the ubiomes of people with a CFS diagnosis. That table is at http://microbiomeprescription.azurewebsites.net/Data/SymptomEndProductExplorer?site=gut&filter=266.

The only items on both lists were:

  • pantothenic acid — both were high
  • Nicotinamide versus niacin were opposite. They are different forms of the same chemical group

Their studies excluded looking at most B vitamins (which raise question on why some B vitamins were included – patients may be taking B-100 supplements).

In terms of B vitamins, the end products predicted from the ubiome results for CFS diagnosis are:

  • LOW Norepinephrine Info 37.79 for significance
    • ” CFS patients had significantly higher levels of plasma norepinephrine” [2016]
  • LOW Dopamine Info 34.41  for significance
    • “Pro-inflammatory cytokines can alter the metabolism of serotonin (5-HT) and dopamine (Felger and Miller, 2012),* effecting dysregulation of associated neurotransmitters, including glutamate, norepinephrine (NE, noradrenalin), and corticosteroids. 5-HT and NE are the major neurotransmitters ” [2017]
  • LOW Tyrosine Info 29.33 for significance
    • ” levels of tyrosine, the rate-limiting dopaminergic precursor, were significantly lower at all time points in the CFS patients.” [2003]
  • LOW Niacin (Vitamin B3) Info  12.41 for significance
  • LOW Succinate Info 5.59 for significance
  • LOW Thiamine (Vitamin B1) Info 4.17 for significance
  • HIGH Folate (Vitamin B9) Info  3.02 for significance
  • LOW Phenylacetic acid Info 2.32
  • HIGH Biotin (Vitamin B7) Info 2.99
  • HIGH Riboflavin (Vitamin B2) Info 2.26
  • LOW Pantothenate (Vitamin B5) Info 2.22
  • LOW Isovaleric acid Info 2.19
  • LOW Cobalamin (Vitamin B12) Info 1.93

Some of these agree with studies and some do not appear to. We need to remember some complicating factors:

  • We are looking at production in the gut, other things may have a hearty appetite there and the production may not make it to the circulating blood.
  • There is the rationing hypothesis — if the body is low in a metabolite, signal are sent out to shut down other consumers, and only essential functions consume it.

Depression

In an earlier post, “Additionally, Isovaleric acid in stool correlates with human depression.[2016] hence isovaleric acid producing bacteria appear to be another facet.”

Looking at those that report depression, http://microbiomeprescription.azurewebsites.net/Data/SymptomEndProductExplorer?site=gut&filter=,287

we found isovaleric acid (LOW), dopamine (LOW) etc. with similar to the above. The problem is that we have depression associated with CFS and true depression mixed in our data.

Bottom Line

This beta-forecast of end-products is interesting because of the number of strong associations found. In many cases, they agree with common accepted knowledge — in most other cases, we find the end product is significant according to the literature but the predicted shift is opposite to what is reported in the literature.

As a data scientist, getting the association is key. The sign of the association is less significant and ceases being an issue if random forest or tensor flow is applied to the data.