Preprint Initial findings from the DecodeME genome-wide association study of myalgic encephalomyelitis/chronic fatigue syndrome, 2025, DecodeMe Collaboration

[ME/CFS Science Blog]

Can you give more detail for how you used to decodeme data to get these lists of traits? Search for genes? Variants? Ranges?

Searched for genes, like CA10. Then you get a list of other traits and GWAS where the gene has been implicated.

Selected the DecodeME genes mainly based on proximity to the SNP with the lowest p-value in the region. The bottom of the table is from regions that only reached a p value of 10^-7.

 

[forestglip]

Searched for genes, like CA10. Then you get a list of other traits and GWAS where the gene has been implicated.

Selected the DecodeME genes mainly based on proximity to the SNP with the lowest p-value in the region. The bottom of the table is from regions that only reached a p value of 10^-7.

Ok thanks. How did you select traits to highlight? For example, BTN3A2 has 97 traits, some with even lower p values than some of the traits in your table, such as height or teeth issues.

 

[ME/CFS Science Blog]

For example, BTN3A2 has 97 traits, some with even lower p values than some of the traits in your table, such as height or teeth issues.

Yeah good point, it's a bit arbitrary. Selected the ones with the most associations (those that you see if you click on 'Traits'). Those are mostly quantitative traits like height or intelligence that have been tested a lot and have big sample sizes.

So I also looked at traits with a reported odds ratio, because those are often binary traits like having an illness or not. Added those binary traits if they had a p-value lower than 5*10^-8. But I didn't try to be very comprehensive so perhaps I missed some.

EDIT: should probably have made clearer that this was only a selection of associated traits. Will update my post above.

 

[jnmaciuch]

@ME/CFS Science Blog I have a selfish question based on a random thought that just occurred to me—did your analysis with other diagnoses happen to include anything for thyroiditis? Interested in both Hashimoto’s or subacute/viral

 

[ME/CFS Science Blog]

@ME/CFS Science Blog I have a selfish question based on a random thought that just occurred to me—did your analysis with other diagnoses happen to include anything for thyroiditis? Interested in both Hashimoto’s or subacute/viral

Not that I can remember but it's quite possible that I overlooked.

 

[jnmaciuch]

Not that I can remember but it's quite possible that I overlooked.

It might not have come up anyways when cross referencing with DecodeME since a lot of risk genes for thyroiditis are on the X chromosome or in MHC. Just wanted to quickly check on the off-chance something had come up already. Thanks for responding!

 

[EndME]

It might not have come up anyways when cross referencing with DecodeME since a lot of risk genes for thyroiditis are on the X chromosome or in MHC. Just wanted to quickly check on the off-chance something had come up already. Thanks for responding!

I don't think we have any information on which conditions were exactly excluded from DecodeME (but could have included people with thyroid autoimmune conditons). Data on comorbid conditions is still forthcoming I believe and the questionnaire included questions on thyroid conditons.

 

[Jonathan Edwards]

I don't think we have any information on which conditions were exactly excluded from DecodeME (but could have included people with thyroid autoimmune conditons).

I doubt that would have skewed the result significantly. None of the known genes for Hashimoto's (e.g. DR, PTPN22) gives a high risk so most people with the risk alleles will still have got into the DecodeME patient cohort.

I amnot aware of X chromosome genes being involved. My guess is that it is the absence of Y that matters for the sex ratio.

 

[hotblack]

Not sure if this has been explicitly asked, but does the lack of MHC appearing to be significant in DecodeME results increase the likelihood of whatever is going on being MHC-independent. In other words, should we be looking more at something involving gamma delta T cells, MAIT cells, NK cells or similar? What other options are there?

 

[V.R.T.]

Not sure if this has been explicitly asked, but does the lack of MHC appearing to be significant in DecodeME results increase the likelihood of whatever is going on being MHC-independent. In other words, should we be looking more at something involving gamma delta T cells, MAIT cells, NK cells or similar? What other options are there?

This is what Chris Ponting said when I asked him about HLA a few weeks ago, thought it might be relevant.

No timeframe for completion of the HLA analysis. The Manhattan plot in the preprint includes all variants including those in the HLA. But note that in the HLA analyses we do not test each DNA variant, rather we test combinations of variants that are commonly coinherited, i.e. "HLA alleles". So even if there isn't a "signal" in the Manhattan plot this does not immediately mean that we won't see association to an HLA allele.

 

[Jonathan Edwards]

Not sure if this has been explicitly asked, but does the lack of MHC appearing to be significant in DecodeME results increase the likelihood of whatever is going on being MHC-independent. In other words, should we be looking more at something involving gamma delta T cells, MAIT cells, NK cells or similar? What other options are there?

You are making the same mistake I made, I think. BTN2A2 is MHC - Class I. There seems to be a DQ skewing as well although DQ is a bit mysterious too. The lack of a DR linkage is a bit against autoantibodies but only a nudge. The lack of HLA-A,B or C means it probably as a spondarthropathy but we new that. I don't think the scope is shifted that much. BTN2A2 hangs around wherever there is antigen presentation.

 

[jnmaciuch]

You are making the same mistake I made, I think. BTN2A2 is MHC - Class I

It is not, it is a separate gene nearby the HLA locus on chromosome 6

 

[hotblack]

BTN2A2 hangs around wherever there is antigen presentation.

Do you mean because it’s a part of a subset of the immunoglobulin family it pops up or is there some other reason or link?

I’d completely forgotten about the HLA-DQ find.

 

[Jonathan Edwards]

It is not, it is a separate gene nearby the HLA locus on chromosome 6

In the MHC Class I region.

Google:
BTN2A2 is a butyrophilin gene located in the Major Histocompatibility Complex (MHC) Class I region, not directly an MHC molecule itself, but rather an MHC-associated gene that plays a role in T cell regulation and immunity.

I don't know what 'not an MHC molecule itself' means. Complement and TNF-related genes are MHC Class 3. They are not HLA but they are 'MHC genes'.

 

[Jonathan Edwards]

Do you mean because it’s a part of a subset of the immunoglobulin family it pops up or is there some other reason or link?

The summaries I have found specifically state that it is found where you find other molecules associated with antigen presentation - including breast epithelium and milk fat globule membrane.

 

[jnmaciuch]

Complement and TNF-related genes are MHC Class 3. They are not HLA but they are 'MHC genes'.

“MHC-III” is a category encompassing all the genes that happen to fall between class I and II complex proteins, including complement, TNF, and a couple dozen other signaling genes many of which have functions far outside immunity, if they can be linked to immunity at all. Same goes for anything in proximity to MHC-I genes. I would know, I spent several months agonizing over ATAC-seq peak calling in this exact region.

When anyone in any of my academic circles speaks of MHC-I molecules or genes, it is explicitly referring to the proteins that make up the MHC-I complex itself. Anything beyond that, you’re expected to talk of the actual genes themselves. Maybe it meant something coherent previously, but several decades of genomics work have probably changed that.

BTN2A2 can be called an “HLA-associated” gene insofar as it has been linked to HLA functionally. Doesn’t mean that must be its only relevant role, just one that’s been written about a lot by people interested in immunity. I’m remembering something you said about reviews—had to do with bibles and li’bles.

 

[jnmaciuch]

I’d completely forgotten about the HLA-DQ find.

It wasn’t mentioned much at all in the manuscript because the DecodeME team had reason to distrust the association.

 

[jnmaciuch]

Not sure if this has been explicitly asked, but does the lack of MHC appearing to be significant in DecodeME results increase the likelihood of whatever is going on being MHC-independent. In other words, should we be looking more at something involving gamma delta T cells, MAIT cells, NK cells or similar? What other options are there?

To try to actually answer your question—as I’ve said in other contexts, it’s hard to exclude anything on the basis of a lack of findings in a genetics studies. That being said, in other diseases where there is a known HLA association, the association has nearly always been strong enough to overcome many of the common pitfalls that lead to results dropping out in genetics studies. So much so that repeat studies for the same illness will often chose to ignore HLA findings because it’s so consistent.

So we can’t conclude that HLA is irrelevant in ME/CFS, but we can also say that the small handful of tentative links found to HLA in DecodeME are of a much different caliber than in other diseases with a much stronger case for HLA involvement.

Personally, because of the uncertainty in even knowing whether differential SNPs were correctly linked to the relevant gene, theories based in one or two weakly associated genes alone don’t hold much water for me. A compelling theory would need to either coherently link together a longer list of genes so you could be confident that at least some of them are accurate annotations, or it would need to have a strong and plausible explanation of mechanism.

 

[Ravn]

In the MHC Class I region.

Google:
BTN2A2 is a butyrophilin gene located in the Major Histocompatibility Complex (MHC) Class I region, not directly an MHC molecule itself, but rather an MHC-associated gene that plays a role in T cell regulation and immunity.

Does this suggest we’re potentially interested in non-classical rather than classical MHC?

Another non-classical MHC Class 1 gene also hanging out on Chr6 is HFE (another tier 1 candidate gene). While not doing classical MHC stuff, it does seem to be involved in antigen presentation in some way as well as in T-cell & NK cell regulation, so in that broad sense a bit like the BTN2A2

 

[Jonathan Edwards]

When anyone in any of my academic circles speaks of MHC-I molecules or genes, it is explicitly referring to the proteins that make up the MHC-I complex itself.

And that is what I had been used to but not so many years ago I realised that MHC refers to the region and that the functional attributions we assume aren't necessarily as simple as we think. Nobody is very clear what the function of DQ is, for instance - hence Danny Altmann's speculative review of it. A variety of atypical interactions occur that don't fit the classic peptide in groove. At some point a region got stamped with the name MHC. It includes gene products with a whole spectrum of related and unrelated functions.

The relevance here is that in the first exposure of the DecodeME data to the advisory board Chris showed a nice peak in MHC. I was puzzled when I saw the published data with a peak for BTN2A2 on chromosome 6 but nothing else. As far as I can see the reason is that BTN2A2 is within MHC.

Just like fundamental physicists, immunologists often use language very badly and not infrequently generate spurious arguments as a result. If they can go around saying a slightly raised CRP is 'systemic inflammation' or build absurd theories on 'TH1 dominance' they are perfectly capable of muddling gene terminology. The review literature shows how confused that actually is.

I got to learn to be very sceptical about how people use words in immunological 'academic circles'. And proved they were all talking horse manure. I advise you to do the same.