Preprint Identification of Novel Reproducible Combinatorial Genetic Risk Factors for [ME] in [DecodeME Cohort] and Commonalities with [LC], 2025, Sardell+

[Jonathan Edwards]

But why assume they all play a part?

Because everything affects everything. A pebble kicked on Chesil Beach contributes to the eruption of Etna, to a tiny extent. We are dealing with quite small causal effects even for the genes Chris has identified. There is no cut off. The bigger the sample you take the more genes will look significant, but since the effects of systematic bias also increase with sample size you end up chasing shadows.

The claim that 259 genes are involved demonstrates a lack of the most basic understanding of the math of the problem.

 

[Sasha]

Because everything affects everything.

I don't think it does...

A pebble kicked on Chesil Beach contributes to the eruption of Etna, to a tiny extent.

The mechanical damping would mean that it wouldn't, surely.

We are dealing with quite small causal effects even for the genes Chris has identified. There is no cut off. The bigger the sample you take the more genes will look significant, but since the effects of systematic bias also increase with sample size you end up chasing shadows.

The claim that 259 genes are involved demonstrates a lack of the most basic understanding of the math of the problem.

I agree that a cut-off is somewhat arbitrary but why would genes for, say, bone growth affect ME/CFS?

 

[forestglip]

I think they named the wrong gene. OLFM4 isn't one of the 259 core genes in this paper's Extended Table 3, as far as I can tell. On the other hand, CSE1L is a DecodeME tier 1 gene that is one of this study's core genes.

And while DCC is one of the PrecisionLife core genes, it wasn't a tier 1 or 2 gene in DecodeME, but the sentence kind of implies it was. It was a gene at a less significant locus in DecodeME.

Will see if I can find a contact method to let them know. [Edit: sent]

So it turns out there's already a new version of the preprint that updated this sentence to say CSE1L instead of OLFM4.

It seems that something is wrong with the MedRxiv website, because it's not showing the "New Version Available" alert when I look at version 1. [Edit: They fixed the website.]

Link to version 2

 

[Jonathan Edwards]

I don't think it does...

I can assure you that physics insists that it does.
Effects may be trivial but where do you draw the line?

I agree that a cut-off is somewhat arbitrary but why would genes for, say, bone growth affect ME/CFS?

For all sorts of reasons. You are probably not aware that genes get used for all sorts of different purposes in different place. The gene that codes for a joint lubricant protein also codes for a megakaryocyte growth factor involved in platelet production. The VCAM-1 gene controls macrophage movement in joints and also is required for B cell survival in bone marrow and lymph node. The calcitonin gene also encodes the calcitonin gene related peptide neurotransmitter.

And a gene for bone growth might take up a certain number of nanometres of chromosome and in doing so contribute to the shape of a histone protein that silenced a synaptic protein gene.

The possibilities are endless.

It is a bit like Microsoft Word. If you type just one letter you may suddenly find your whole document is re-formatted for mysterious reasons only known to Bill Gates. Genes are very much like that.

 

[Sasha]

For all sorts of reasons. You are probably not aware that genes get used for all sorts of different purposes in different place.

Interesting! Thank you.

 

[Michelle]

For this they were giving me heparin. For two days I did not have any ME symptoms. Then after 2 days the problems began again. I will never forget it. Took me years to understand the connection that there was something with Heparin.

Note also that CA10 (DecodeME) is associated with Heparan Sulfate. We are definitely on the right track, I will look further to your results. Thank you for this work.

Off topic for this thread, I know, but this is yet another potential heparin responder.