Genome-wide association analysis identifies genetic variations in subjects with ME/CFS

[Valentijn]

@Valentijn , apparently I have more trust in our doctor, KDM. Do I get you right that you dismiss this study, or is there anything of value in it, in your opinion?

He's a great doctor, but faith in the researchers involved has little to do with the research and the science. I'm not dismissing it completely ... it may have found one or two bits of value. But it's not going to be the basis for diagnosing ME/CFS, or predicting who will get it.

At the very least, the relevant results need to replicated in a new set of patients.

 

[Helen]

I'm not dismissing it ... it may have found one or two bits of value. But it's not going to be the basis for diagnosing ME/CFS, or predicting who will get it.

Time will tell.

At the very least, the relevant results need to replicated in a new set of patients.

The group of researchers told they expect that to be done too.

I can´t see that if we, diagnosed with ME/CFS, compared our available SNP´s with the results in their study would be less interesting than when you check data in the 23andme tests files that some us have sent to you. I regret that we couldn´t have a more open-minded discussion, but I assume you don´t think it deserves that. I will ask KDM about comments on your critical comments during next appointment as I find the study valuable and a big step ahead in search for the cause of ME/CFS.

 

[Valentijn]

I regret that we couldn´t have a more open-minded discussion, but I assume you don´t think it deserves that.

I don't think I ever said that, nor discouraged anyone from sharing their results for the SNPs they have data for. But believing that another SNP study is substantially changing the knowledge, perception, or treatment of ME/CFS is simply incorrect.

If you disagree with any of the points I made in that regard in my earlier post, perhaps you could specify them?

 

[msf]

I think the main reason this paper isn´t generating the same kind of discussion we had over other papers is that no one knows enough about this subject to be able to say whether the findings are significant or not. If someone can tell me what effect mutations in the coding sections of T-cell receptor genes might have, I would love to hear from them!

 

[voner]

@msf, @Helen, I find the lack of discussion rather unfortunate also. if nothing else, if any of you are a patient of Dr. DeMierlier, it be interesting to hear what his speculation is on what they found. i'd love to hear his take (Or anyone else's at this point) on the patient subgrouping that they show in figures 2 and 3.

@Helen, – I have no problems with you breaking off this thread with another discussion. I ran my saliva through 23andme in 2015 and that version of their chip did not check any of the snps cited in the paper.

 

[Simon]

He's a great doctor, but faith in the researchers involved has little to do with the research and the science. I'm not dismissing it completely ... it may have found one or two bits of value. But it's not going to be the basis for diagnosing ME/CFS, or predicting who will get it.

At the very least, the relevant results need to replicated in a new set of patients.

Well, here's the view of an independent expert on genomics, George Davey Smith, who spoke at the recent CMRC conference. In the past he's criticised the whole way gene asssociation studies have been overinterpreted: not someone who's afraid to speak his mind:

https://twitter.com/i/web/status/698117629404450816

I'm about to blog about his talk - part one later today, with a bit of luck

So I think genetic studies need to take a much more systematic approach, by focusing on the historically more relevant sections of DNA, such as exons and splice sites.

Maybe, but with the function of so much of the genome unknown, it's a risk to dismiss non-coding sections. Certainly genetic changes in introns and gene promoters can make a difference, as well as genetic differences quite a long way from genes. That's before including the impact of things like regulatory long-non-coding-RNAs.

One way to try to assess the biological relevance of SNPs is to look at gene expression in different tissues (inc the brain). There is a new post-mortem study that does this - more in part 2 of my blog.
GTEx detects genetic effects | Science

 

[msf]

For the stat-stupid, what would a replication sample consist of in this context?

 

[msf]

I wonder why Davey-Smith didn´t point out that there´s not much info given on the patients either.

 

[Valentijn]

Well, here's the view of an independent expert on genomics, George Davey Smith, who spoke at the recent CMRC conference.

Maybe a little harsh. The paper seemed pretty typical of SNP studies ... unfortunately the standard is a bit low in that area. Probably due to a lot of people finding it a cheap and easy way to look for something, without a strong background in genetics or statistics.

Maybe, but with the function of so much of the genome unknown, it's a risk to dismiss non-coding sections.

I sort of agree. If cost is a concern (it always is), studying the exome is a lot more likely to generate relevant results for a lot less money. These types of almost-random all-over-the-place SNP sampling chips are even cheaper ... but I think those cheap costs are almost completely wasted in research since it's not an intelligent sampling of SNPs.

One way to try to assess the biological relevance of SNPs is to look at gene expression in different tissues (inc the brain).

Yeah, but that requires genetics experts with good equipment and funding. The vast majority of the type of SNP research in this study is being done by non-geneticists. And when there's so much SNP spam research, the real genetics researchers are probably not even interested in reading most of it, much less following up on it in a more rigorous manner.

 

[Valentijn]

For the stat-stupid, what would a replication sample consist of in this context?

You'd have two groups of patients. The first group would be used to see if the alleles of some SNPs are more or less frequent compared to controls. Then the 2nd group of patients is used to see if the results from the first group can be confirmed.

This can be done in the same study, or in a 2nd study. It doesn't seem to happen often in the same study, maybe because it's more desirable to create two publications instead of one. Or if the replication sample doesn't confirm the first finding, you at least still have one positive paper getting published, and a 2nd which can possibly be buried.

When the replication group is in the original study, I think the ones I've seen have had a sample size of about half of the original patient group. I don't think they had nearly enough patients to use some for replication in this study.

 

[msf]

That´s what thought it would be, but I don´t get why that would help (stat-stupid!). How is having two groups better than having 80 patients in one group?

 

[msf]

Also, is there any specific reason a genetic study in particular should have a replication sample in it? As Valentjin pointed out, this isn´t ubiquitious in studies of other aspects of disease.

 

[Valentijn]

That´s what thought it would be, but I don´t get why that would help (stat-stupid!). How is having two groups better than having 80 patients in one group?

Replication is a big step in filtering out the false-positives. Including this in the original paper, instead of waiting to do it some years down the road, is a definite bonus.

Instead of comparing 900,000 SNPs between controls and patients, the replication group is being compared based on the 400 SNPs found with the initial group. So it becomes easier to show statistical significance, since they're no longer making such a huge amount of comparisons.

 

[Simon]

Maybe a little harsh. The paper seemed pretty typical of SNP studies ... unfortunately the standard is a bit low in that area. Probably due to a lot of people finding it a cheap and easy way to look for something, without a strong background in genetics or statistics.

Not really - that's the problem with some of the field, but not all

And if I may gratuitously plug my blog
The Power and Pitfalls of Omics: George Davey Smith’s storming talk at ME/CFS conference

He began his CMRC talk by showing a photo of a boy in a dunce’s hat and declaring, ‘I know nothing about this illness’. But he does know about genomics research, and on that basis, he started off in a typical style for him: breaking a few eggs in pursuit of the right answer.

He showed four recent ME/CFS studies identifying links between differences in some genes and the illness. Davey Smith was not impressed.

‘The statistical power [to detect a real effect] is literally zero’, he said. These studies were simply too small to show anything, and any apparent findings are effectively guaranteed to be false positives — that is, associations simply happening by chance.

‘ME/CFS gene studies today’, said Davey Smith, ‘appear much like other gene association studies a decade ago — hopelessly unreliable.’

Davey Smith and his colleagues helped to dramatically reduce such problems more than a decade ago. They wrote a paper for The Lancet ‘that didn’t make us enormously popular’, pointing out that almost all published association studies up until 2002, including those published in The Lancet, were proving unreliable.

They argued that the false associations were showing up primarily because of publication bias (only studies that found an association got published, while those that did not were ignored), way-too-small sample sizes, and poor statistical techniques.

As a result of that 2003 paper, things changed. Funders such as the Wellcome Trust took the lead and refused to finance further studies unless researchers collaborated to create studies that were big enough to give reliable results.

The upshot was that huge numbers of genetic variants discovered since 2005 have stood the test of time, while almost all of the associations people found before ‘have now just gone’.

Researchers can now search through more than 10 thousand robustly established genetic associations with disease, including obesity, diabetes and heart disease. But currently there are none for ME/CFS.

The Grand Challenge is poised to change that.

So it is possible to do robust GWAS studies, they just don't look like any mecfs ones done to date, and that needs to change.

 

[Valentijn]

Not really - that's the problem with some of the field, but not all

Nearly all Based on what I see in current non-ME/CFS SNP research papers, I disagree with the conclusion that most of it got cleaned up 10 years ago.

 

[msf]

Replication is a big step in filtering out the false-positives. Including this in the original paper, instead of waiting to do it some years down the road, is a definite bonus.

Instead of comparing 900,000 SNPs between controls and patients, the replication group is being compared based on the 400 SNPs found with the initial group. So it becomes easier to show statistical significance, since they're no longer making such a huge amount of comparisons.

Ah, got it, thanks.

 

[msf]

I have a couple of questions for Davey Smith, if you are in touch with him, Simon. Why doesn´t he do a study he regards as good? Or support/fundraise for such a study?

Edit: I see from your article that he is planning to help with a study. Good for him!

 

[Sea]

But Davey Smith stressed that we learn ‘nothing at all if the sample size is too small’. Many thousands are needed as a minimum. The Grand Challenge, aiming to collect data from 10,000 patients, is the first time such a thing has been attempted for this disease.

I think this is relevant from Simon's blog

 

[beaker]

In my opinion this study deserves a lot more of attention from PWME. I think it mediates hope and future treatment possibilities.

Surprisingly a study that indicates genetics as a cause of ME/CFS was published in a well-reputed journal of psychiatry. It´s time for the psychobabblers to change side; ME/CFS IS a disease or a syndrome of physical origin.

emphasis mine.

Genetic studies are interesting and are important for many reasons. But I just feel the need to point out that genetic cause does not explain cluster outbreaks.

 

[msf]

I also notice that everyone is still discussing this study in the abstract. It would be good if Davey Smith could answer a specific question, such as: if you assume that these are false associations, what is the probability that 6/12 ´statistically significant SNPs´ found in the coding regions of genes would be found in the coding region of genes that have a role in the immune system? Of course, it would be impossible to determine the answer to a high degree of accuracy, but a rough calculation should show whether it is quite likely/very unlikely/incredibly unlikely etc.

To do this calculation, it seems to me, you would have to identify how many SNPs in the The Affymetrix Genome-Wide Human SNP Array 6.0 (Affymetrix) are in the coding section of genes that are known to have a role in the immune system. You would then need to do some statistics...so over to you guys! May the best statistician win!