HLA and pathogens in myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and other post-infection conditions, 2025, Georgopoulos et al

[Jonathan Edwards]

Could you possibly clarify what you mean by this for laypeople like myself?

I don't even remember the detail but I think it may have been related to the way HLA allele rates can vary very easily between populations even if matched ethnically. I have discussed this with friends who work on DQ. The answer was that it is the biggest can of worms of all because there are so may variants and these vary subtly in rates between ancestries and populations. So it is the easiest thing of all to think you have a DQ link when you haven't.

Is DR the HLA that DQ is usually linked to but wasnt in DecodeME?

No I think it was the gene for the other DQ chain. If I remember right DQ is pesky because both its chains vary so you have a combinatorial number of variations of the binding groove with one chain on each side.

 

[Simon M]

I don't even remember the detail but I think it may have been related to the way HLA allele rates can vary very easily between populations even if matched ethnically.

Yes, and it's a nightmare to untangle. The DecodeME analysis restricted itself to people with the white European ancestry, which simplifies things – but not a lot.

One reason why HLA analysis is so challenging – population stratification

HLA varies enormously even in White Europeans, presumably because – at different times in our history – different HLA all have protected against e.g. pandemics, and that gives you a strong selection effect, and so creates strong differences between population groups . Especially as, in the past, people were less mobile so pandemics were on a smaller scale, with less population mixing and different groups had different HLA selective pressure.

If you're white British, you have a colourful backstory. We've been invaded or conquered by Romans, Saxons, Celts,Gauls, Normans, Vikings and no doubt others I have forgotten. Plus more recent European immigration. And quite a lot of Irish immigration (thanks, Dad). And there will have been plenty pre-Roman action as well.

A good measure of how much of a problem this is genomic inflation (see Kitty's explanation a couple of posts down). In the final DecodeME analysis, this was something like 1.066, which is close to 1 andpretty healthy. In earlier analyses, it was higher, though identifying how to minimise the population stratification to controlled genomic inflation isn't easy. It helps that the project was set up with UK biobank as controls – it has half a million in its cohort - making tight control matching easier.. Just over half of UKB individuals were included as controls in the final analysis.

Population stratification is likely to be a big issue in other genetic studies that don't sufficiently match controls on ancestry.

DecodeME used a different, more detailed approach for its preprint HLA analysis. I think the new analysis which will be in collaboration with someone who's an expert in this area, will be different again.

Sorry, I got a bit carried away then. Congratulations, you made it to the end.

 

[Kitty]

Thanks, @Simon M, that's really helpful.

I couldn't make sense of one sentence because I'd never heard of genomic inflation, so here's a Google explanation for anyone else who hasn't:

Genomic inflation is a statistical phenomenon where test statistics in a genome-wide association study (GWAS) are systematically higher than expected, often due to technical or biological factors like population structure, non-random sampling, or a trait being influenced by many small genetic effects (polygenicity). It is quantified by the genomic inflation factor (λ)*, which is calculated by comparing the median of the observed test statistics to the expected median. A λ value greater than 1 indicates that the results are inflated, and methods like genomic control are used to adjust for this inflation by dividing the observed statistics by λ.

Your voice software has misread one word (buyer instead of bio), which also threw my fogged brain for a second or two!

It helps that the project was set up with UK buyer bank as controls