Genetic Risk Factors for ME/CFS Identified using Combinatorial Analysis, 2022, Das et al

I think this makes sense. To me ME is not a linear process , it's a systems condition ,which reminds me of climate science .
It's a "both and " condition not a binary one.
This is difficult for medicine in particular to get its head around

We also don't know everything each gene does as they can multitask- particularly important for signalling so feedback loops are important . So an association of genes seems to me a good idea. It could give a hint for a clue for mechanisms and subgroups .

to me this makes sense.

It also may give insights into repurposing drugs.

This group have been involved with motor neurone (ALS) research and some of the drug tests on cell assays , based on genetics, have seemingly been encouraging .

 

Great graphic . Explains things well to a non science person

 

Looks good @Simon M , the visual makes it easier to grasp the concept.

 

I found your written explanation clear but with the graphic my brain insists - wrongly, I know - that the grey circles inside the square and inside the circle belong to the square and circle subgroup respectively.
Would it work to have all the grey circles outside the square and circle but still have the first picture look 'messy' enough? Or alternatively have them form a third subgroup, say a triangle (you mentioned 3 subgroups in the text)?

 

I've now published a blog, thread https://www.s4me.info/threads/blog-new-study-links-14-genes-to-me-cfs.29511/#post-437195

Thanks for the feedback on the graphics I posted. It was clear I need to make some changes to keep everything clear, I'm afraid I ran out of energy. I might add them later.

 

I am not sure how well your diagram illustrates the maths behind the problem intuitively, @Simon M. It is hard to think of a way to do it but what about this:

The subgroup hypothesis implies that the cohort can be sorted according to several independent variables or 'dimensions'. You could plot each gene on a different axis of a graph with xyzabcdefghij.. axes, except of course that such 'brane' structures that string theorists use are incomprehensible. But there might be a simpler metaphor. Consider a plantation of tall thin redwood trees with four different colours of bark. If you walk around the plantation at a distance with binoculars you will make out coloured trunks shifting past each other as you walk, apparently all separated by arbitrary amounts and intermingled. However, if you walk in to the plantation you may find that an increasing number of trees to the north east are red and more to the north west are dark brown. Eventually you reach a point where all trees are seen in perfect rows, with all trees north east red, all to the north west dark brown, all to the south east streaked and all to the south west blotched. The trees are actually planted in strict rows with NE, NW, SE and SW sectors of a different colour. The trick is to sort by the right axis. In genetic terms there are disease subgroups linked to having a north and an east gene or a south and a west gene but it is hard to see the grouping without plotting them out according to the genes. Once you have done that it may become clear that all in one group have certain features and those in another a different set of features etc.

 

Thanks, it's a very interesting idea, but you may have overestimated by PowerPoint skills.

What the paper doesn't show is the link between subgroups and genes. Or even clear description of subgroups e.g. (hypothetically) infectious onset in teens, POTS, more severe - because there is limited data in the Biobank. DecodeME will provide richer data - at least on symptoms and co-morbidities.

My view on genes is that we should be cautious, not least because the results didn't replicate well. I'd have more confidence once they have DecodeME data (assuming data access is agreed). This will give more accurately diagnosed subjects, as well as a larger sample. Although they still probably won't want analyse more than a few thousand subjects, they can, for instance, take 6000 subjects, randomly divided in half and analyse the two groups separately. I think this would give the best chance of a meaningful replication.

 

Link with Long Covid & MS - work underway
I didn't include this in my blog because of my concerns about the accuracy of gene identification I mention above, but I still find it very interesting.

Using a hypothesis-free combinatorial analytics approach based on the PrecisionLife platform, we identified
199 SNPs in 84 high-order combinations that were highly associated with 91% of the ME/CFS cases in the UK
Biobank Pain Questionnaire cohort. These variants could be mapped to 14 genes, which appear to be
compatible with the major cellular mechanisms suspected by other groups working in the field and show a
level of overlap with diseases sharing similar symptoms, such as MS111and long Covid(112).
...

Similarities with other Diseases
MS and ME/CFS patients share a number of similar symptoms, including pain, sleep disturbance and
cognitive dysfunction117, and both can have a viral trigger such as Epstein-Barr virus (EBV)4,118. There is also
increasing evidence that many patients diagnosed with long COVID share similar symptoms, such as chronic
fatigue and ‘brain fog’, with individuals with ME/CFS. It is also believed that some patients may be developing
ME/CFS as a direct result of having a COVID-19 infection119,120,121.

This suggests that the two diseases may share similar etiologies with possible overlap in the biological
drivers and risk genes. Our analysis of the first UK Biobank COVID-19 population identified four genes out of
68 associated specifically with the risk of severe COVID that we had previously identified as having strong
association with neurodegenerative processes23, including ATXN1, SORCS2 and STH and MAPT from loci on
chromosome 17 that were subsequently validated by the results from the COVID-19 Host Genetics
Initiative122. This analysis also revealed several other disease and symptom associated mechanisms, such as
viral host response factors and pro-inflammatory cytokine production.

We are in the process of analyzing two populations in long COVID-19 (Sano Genetics, GOLD study) and
multiple sclerosis (UK Biobank) in order to identify any shared genes and biological mechanisms underpinning
ME/CFS, multiple sclerosis and long COVID-19 development. Preliminary findings from our long COVID
analysis have indicated that three of the genes identified in this study are also significant in the long COVID
patient group (albeit with different SNPs, but again none of these are in LD). These will be subject of further
validation in a new publication later this year

 

Main subgroups within key genes were metabolic , host response, autoimmune and sleep

 

coul you please explain the difference btween linear and non-linear process? I can't get to understand it... thank you

 

I'll have a think and see if I can post something that makes sense , perhaps with examples tomorrow .
It may be if course that what I think is non linear is not what others think !

 

In very simple terms linear processes are those in which what you get out is proportional to what you put in, without any feedback or interaction effects.

So the number of mail vans going around carrying letters might be proportional to the number of people in a city - because the number off letters is on average the same for city inhabitant groups. Linear processes can also usually be added and still give you a linear result. So the number of heavy lorries for parcels plus the number of mail vans is still proportional to the number of people in a city.

But things would get non-linear if above a certain threshold cities found it more efficient to put all the mail in heavy lorries, or had more people too poor to send big parcels or with plenty of big stores there was less need to send some things by mail,

Feedback loops make things non linear. Thermostats and homeostatic endocrine systems are typical examples - of stable nonlinearity. For quite a lot of diseases positive feedback loops set up - including autoimmunity, cancer and some kidney disease - which gives unstable nonlinearity.

The interesting question is whether or not we expect the effects of gene variants to interact in the way they confer risk. Maybe a variant of gene x increases risk by 1.2 and a variant of gene y increases it by 1.2 but the two variants together increase it by 6.0. The same might be true of variants of genes p and q but it might be that variants of x and q increase risk by the expected 1.44 or even maybe less.

I can think of lots of hypothetical mechanisms but I don't actually know of any documented ones. That may reflect the difficulty in identifying such interacting effects.

 

thank you for explaining, @Jonathan Edwards, that is very kind.

 

Great explanation thank you .
Lorenz's butterfly effect is probably the best idea I had - it illustrates how disproportionate the size of input to effect can be and how many variables with no direct connections can be affected

 

it's a good point, but I don't think so. In fact, some of the ME patients also reported an MS diagnosis. I asked the researchers if this could explain the findings. They went away and checked the numbers and said no because there were few cases.

Added:
Also, in diagnostic accuracy studies of this disease, MS is a rare misdiagnosis.

 

I don't pretend to understand the science here at all, but one simple analogy occurred to me that might help illustrate the significance of looking for combinations of SNPs.

If you think about culinary ingredients, there are a huge number of potential ingredients, and a great many dishes will share the same ingredients with each other. So if you were to look at many different culinary dishes, and seek to identify which dishes have what ingredient in, and do that for each ingredient, you would have a huge number of hits with very considerable overlaps. If, however, for each dish, you were to look at each one's combination of ingredients, the number of hits you would get would be greatly reduced, as would the number of overlaps.

I appreciate this analogy is likely pretty woolly, and doesn't account for the fact that various dishes can have near identical ingredients, but vary considerably according to the preparation and cooking process. But I thought I'd put the notion up anyway.

 

thank you!