New techniques — and monotonic thinking

I just read “A metagenomic approach to investigate the microbial causes of myalgic encephalomyelitis/chronic fatigue syndrome: moving beyond XMRV” on About.

While I have strong agreement with this on first read, a second read revealed some subtle but important issues.

“Nevertheless, even if the causal microbe is no longer present, it may have changed the composition of the microbiome, by altering the presence or relative abundance of other microbes. Such changes could be detected by metagenomics and be used to diagnose and, in principle, treat the symptoms of ME/CFS, even if not the initial cause.”

My conclusion from my own review of the literature, is that the causal microbe is either no longer present, or no longer in an acute (active) state. I also agree with the alteration of the micro-biome suggested.

Where I differ comes from my statistical training and work doing machine learning at Amazon. Given the huge range of families, species and strains in our micro-biome, the sample size needed to get any significant (and reliable, reproducible) results is likely 400,000 CFS patients with 400,000 controls. The testing costs to collect this data would be around $200,000,000.

Even once this is done, the question arises on how to correct this when there is estimated to be at least 10,000 species (and likely 200,000 strains!).  Anything impacting one strain may impact hundreds of other strains.

The author being cited has a feeling of expecting to find a single species and not complex interdependencies of many strains that as a whole causes the problem.

The philosophy is right, the requirements for a successful execution is far more challenging then imagined….