Ion Channel SNP Paper

[Valentijn]

They used Fukuda, with no apparent requirement for participants to have PEM.

23andMe tested for rs11142508, rs1160742, rs1328153, rs3763619, rs7865858, rs1504401, rs2383844, rs4738202, rs6650469, and rs655207 on the V3 (previous) chip.

Here's how their results compare on those SNPs to the data of 31 ME forum patients and ethnically matched controls I have full data for:

rs11142508 (C) has 40.3% allele frequency for our ME patients, and 35.5% allele frequency for our controls. It has 55.6% prevalence in the general population, but 35-40% in European populations.

rs1160742 (A) has 37.1% allele frequency for patients, and 35.5% for controls. It's at 50.32% in the general population, and 35-45% in Europeans.

rs1328153 (G) has 6.5% in patients, and 14.5% for controls. It's at 24.2% in the general population and 20-25% in Europeans. This is opposite to the results of the study.

rs3763619 (T) has 38.7% in patients, and 35.5% for controls. It's at 54.6% in the general population, and 30-40% in Europeans. It's probably pretty tightly linked to rs11142508.

rs7865858 (A) has 40.3% in patients, and 37.1% in controls. It's at 39.1 in the general population.

rs1504401 (T) has 14.5% in patients, and 9.7% in controls. It's at 19.9% in the general population and 10-15% in Europeans. This is opposite to the results of the study.

rs2383844 (G) has 32.2% in patients, and 35.5% in controls. It's at 37.8% in the general population.

rs4738202 (A) has 24.2% in patients, and 24.2% in controls. It's at 24.0% in the general population.

rs6650469 (T) has 46.8% in patients, and 48.4% in controls. It's at 40.0% in the general population.

rs655207 (G) has 46.8% in patients, and 45.2% in controls. It's at 38.8% in the general population.

Basically none of the results they reported are echoed in the data from ME patients here. Some had the opposite trend, some were identical or nearly identical for patients versus controls, and many were tightly linked so essentially duplicates. And most were common as dirt.

So nothing interesting at all, as far as I can see.

 

[Jonathan Edwards]

Interesting. I have been working on the idea that what we most need to ME is a data repository of all results for all research so that positive and negative (including unpublished) replications are all lined up for everyone to see.

I now realise that maybe we already have this repository. It is called Simon and Valentijn, not to mention a number of others. Whenever a question arises it seems that someone has all the old data on tap. But I wonder if it exist in a formal repository of any sort - a sort of Cochrane meta-analysis of research database. If not maybe it should?

 

[user9876]

Interesting. I have been working on the idea that what we most need to ME is a data repository of all results for all research so that positive and negative (including unpublished) replications are all lined up for everyone to see.

I now realise that maybe we already have this repository. It is called Simon and Valentijn, not to mention a number of others. Whenever a question arises it seems that someone has all the old data on tap. But I wonder if it exist in a formal repository of any sort - a sort of Cochrane meta-analysis of research database. If not maybe it should?

The whole paper reads to me as if it were in a foreign language. My level of biology isn't good; I might try and read around the details over the next week or so so I can understand.

From a stats perspective am I right in saying that the SNP bits are either turned on or off and the frequency counts in the table represent the proportion of patients where a SNP is set? In which case I don't really see how the table tells us much. It would be great to get the data and look for clusters etc; I quite like the network analysis that Hornig and Lipkin used to recognise connections between measures. Basically the question is are there clear clusters (over the multidimensional data) that can be labelled patient or control or both. If my assumptions about the form of the data are correct (which I have doubts about) then I would start by looking at data using something like a hierarchical cluster analysis and probably a few other clustering techniques.

So I think we do need a data repository so we can look at data in different ways.

 

[Snow Leopard]

I now realise that maybe we already have this repository. It is called Simon and Valentijn, not to mention a number of others. Whenever a question arises it seems that someone has all the old data on tap. But I wonder if it exist in a formal repository of any sort - a sort of Cochrane meta-analysis of research database. If not maybe it should?

At best, we have the summary of papers listed here:
http://www.meresearch.org.uk/information/research-database/

I floated the idea of creating a searchable/mappable database of findings, to a number of forum members, but it didn't really gain traction as I have no idea what software to use to manage it, or to manage the people volunteering their time for such a project. Or basically, it seemed like a lot of work, with dubious benefit. One of the initial goals was to get feedback from researchers, especially young researchers on the benefits of such and how it should be constructed for usability. (and look to see if there are any examples for other diseases too!?!)

There are lots of interesting (broad based) ways of looking at data from papers now, for example:
http://evexdb.org/

I'm wondering what your opinions of creating a searchable research summary database, in some sort of wiki format.

Eg category, (possibly multiple categories, eg "gene expression" along with "cytokine" or "kinase" or whatever, and a simple test vs result format.

Ideally it could be done in a database sort of way, eg click to list by "gene expression to get a list of gene expression results form all the different studies, listable by the different headings. Or then click on "kinase" and get a page of all studies with findings associated with kinases, including any gene expression results.

The key is not having to write up the different tables multiple times like in a wiki article, while at the same time providing opportunities to very quickly build a mental map of what is going on, without reading 50 papers in a day to get there.
A separate part of the wiki could be organised to propose and discuss various hypotheses.

So you guys are my first form of peer review. Does this sound like a good idea (in terms of actually making life easier for researchers)?

What software could be used to manage this?

How should this be organised/managed (people wise)?

 

[adreno]

@Snow Leopard,

Would something simple like Mendeley work?

https://www.mendeley.com/

 

[Jonathan Edwards]

The whole paper reads to me as if it were in a foreign language. My level of biology isn't good; I might try and read around the details over the next week or so so I can understand.

From a stats perspective am I right in saying that the SNP bits are either turned on or off and the frequency counts in the table represent the proportion of patients where a SNP is set? In which case I don't really see how the table tells us much. It would be great to get the data and look for clusters etc; I quite like the network analysis that Hornig and Lipkin used to recognise connections between measures. Basically the question is are there clear clusters (over the multidimensional data) that can be labelled patient or control or both. If my assumptions about the form of the data are correct (which I have doubts about) then I would start by looking at data using something like a hierarchical cluster analysis and probably a few other clustering techniques.

So I think we do need a data repository so we can look at data in different ways.

As I see it each SNP is an indicator of a variant of the gene for a protein that might encode a form of the protein with a slightly stronger or weaker (or more or less specific etc.) function. So it is a bit like if you have SNP 012345678 then that means that you make a variant of a protein (in common with some proportion of the population and not others) that might be a little bit e.g. 'stronger' than in the other people. If this affects a threshold for tripping a feedback loop then it can confer disease risk. The same feedback loop might be tripped by all sorts of variants in either the same protein or one of a family of similar proteins or even an unrelated protein. By analogy with a fuel that makes a car backfire, you might be predisposed to this if you had a SNP that goes with a particular octane rating for the petrol or a SNP that goes with not enough lead additive, etc.

The implication is that all the SNPs with higher frequencies in the patients might go with gene variants that encode for loop-tripping variant proteins. You might then ask what one would expect to find in terms of combinations of SNPs in individuals. I think things get very complicated here because some SNPs go around linked together, as Valentijn points out. Moreover, since a loop trip is a loop trip then there may be no particular increase in people with a double dose of loop tripping SNPs - or there might be. But it is doubtful that the data would be firm enough to start testing that..

I am not sure whether that will make much more sense than the paper but hopefully it will not just confuse more!

 

[snowathlete]

Thanks

Good news: just waiting for final confirmation from Prof Sonya Marshall-Gradisnik, but it appears the authors DID correct for multiple comparisons - and there are more papers to come on this work. Which means we can all get on with discussing this paper's findings.

Thanks for reporting back Simon. That's good, so it means I bothered to read the rest of this thread.

They used Fukuda, with no apparent requirement for participants to have PEM.

23andMe tested for rs11142508, rs1160742, rs1328153, rs3763619, rs7865858, rs1504401, rs2383844, rs4738202, rs6650469, and rs655207 on the V3 (previous) chip.

Here's how their results compare on those SNPs to the data of 31 ME forum patients and ethnically matched controls I have full data for:

rs11142508 (C) has 40.3% allele frequency for our ME patients, and 35.5% allele frequency for our controls. It has 55.6% prevalence in the general population, but 35-40% in European populations.

rs1160742 (A) has 37.1% allele frequency for patients, and 35.5% for controls. It's at 50.32% in the general population, and 35-45% in Europeans.

rs1328153 (G) has 6.5% in patients, and 14.5% for controls. It's at 24.2% in the general population and 20-25% in Europeans. This is opposite to the results of the study.

rs3763619 (T) has 38.7% in patients, and 35.5% for controls. It's at 54.6% in the general population, and 30-40% in Europeans. It's probably pretty tightly linked to rs11142508.

rs7865858 (A) has 40.3% in patients, and 37.1% in controls. It's at 39.1 in the general population.

rs1504401 (T) has 14.5% in patients, and 9.7% in controls. It's at 19.9% in the general population and 10-15% in Europeans. This is opposite to the results of the study.

rs2383844 (G) has 32.2% in patients, and 35.5% in controls. It's at 37.8% in the general population.

rs4738202 (A) has 24.2% in patients, and 24.2% in controls. It's at 24.0% in the general population.

rs6650469 (T) has 46.8% in patients, and 48.4% in controls. It's at 40.0% in the general population.

rs655207 (G) has 46.8% in patients, and 45.2% in controls. It's at 38.8% in the general population.

Basically none of the results they reported are echoed in the data from ME patients here. Some had the opposite trend, some were identical or nearly identical, and many were tightly linked so essentially duplicates. And most were common as dirt.

So nothing interesting at all, as far as I can see.

But, this list looks underwelming to me. Without doing any real math, my brain tells me just from scanning it briefly that there is way too much chance involved here.

As I have 23andme with the SNPtips plugin it's easy for me to see my results in comparison and I am almost entirely the opposite - I think Valentijn has my data too so I'll be one of the ones mentioned already. The most interesting thing for me is that I am so the opposite; it seems statistically unlikely. But that is the problem with genetics studies in most diseases, that random chance causes problems. Toss a coin ten times and statistically it should come up heads 5 times, but it often does not. Although it's unlikely you can get eight, nine or ten heads, which seems significant, but it is still just chance.

The problem though is that a genuine genetic difference in patients might get dismissed on this basis (because it looks like it could be chance). So I try to go easy on genetics papers for that reason. But unless there is an obvious and likely link to disease mechanism -- which there isn't here -- the only way to find out if it is a genuine finding is to increase the sample size to reduce the liklihood of chance, and see what you get.

Personally, with funding for ME/CFS reseach as it is, if I were researching ME/CFS I would probably put my research effort into something other than genetics, unless I had some other reason to suspect a specific mechanism at work that justified looking at the genes involved in that mechanism. But maybe these researchers do?

 

[nandixon]

As I see it each SNP is an indicator of a variant of the gene for a protein that might encode a form of the protein with a slightly stronger or weaker (or more or less specific etc.) function. So it is a bit like if you have SNP 012345678 then that means that you make a variant of a protein (in common with some proportion of the population and not others) that might be a little bit e.g. 'stronger' than in the other people.

Just to mention that in addition to SNPs that affect the function of a protein/enzyme, such as what you're referring to, and which are found in the exon portions of genes (and which code to make the protein), there are other non-coding SNPs on each gene that are potentially just as important and which affect not the function of the resultant protein, but rather how much protein is made and/or at what rate and/or under what circumstances. So it's sort of a quality (how well the protein/enzyme functions) versus quantity (how much and when the protein/enzyme is made) issue. And that's a gross over-simplification.

 

[deleder2k]

Researchers shed new light on cause of Chronic Fatigue Syndrome:

Published on MedicalXpress.com May 11

New research findings may shed new light on the potential cause of Chronic Fatigue Syndrome/Myalgic Encephalomyelitis (CFS/ME).

Researchers from Griffith University's National Centre for Neuroimmunology and Emerging Diseases (NCNED) - part of the new Menzies Health Institute Queensland - have uncovered significant factors contributing to the pathology of this illness.

The results reveal genetic changes in important receptors associated with immunological and cellular function and contribute to the development of this complex illness.

"These findings have been achieved through a team effort involving researchers, patients, funding bodies, clinicians and the support of Griffith University and the Queensland Government," say chief investigators Professor Sonya Marshall-Gradisnik and Professor Donald Staines.

Co-researcher and consultant immunologist Professor Pete Smith said that important signalling mechanisms are disrupted as a result of these genetic changes involving the detection and response to threats.

"These are primitive genes that are involved in many cellular signals in the brain, gut, cardiovascular and immune systems, as well as in the mediation of pain."

These research findings coincide with International Neuroimmune Awareness week commencing Monday 11 May.

The Griffith Health Centre on the university's Gold Coast campus is being lit up each evening from 10 -12 May to raise awareness of neurological conditions such as CFS/ME as well as other conditions such as Fibromyalgia and Gulf War Syndrome.

"The lighting up of the Griffith Health Centre signifies Griffith's commitment to the CFS patient community and our team approach to this research," says Pro-Vice Chancellor (Health) Professor Allan Cripps.

CFS/ME is a highly debilitating disorder characterised by profound fatigue, muscle and joint pain, cerebral symptoms of impaired memory and concentration, impaired cardiovascular function, gut disorder and sensory dysfunction such as noise intolerance and balance disturbance. Many cases can continue for months or years. It is believed to affect around 250,000 Australians.

The research findings are to be presented at an international conference in London later this month.

 

[alex3619]

there is way too much chance involved here.

The problem with false positives is a major issue. At best all findings are candidate findings for further testing. However, invoking the numbers game here, if there are huge numbers of potential cadidates there is a chance that some of them will be really important. The science is about finding out which of them are important, and which are not.

 

[alex3619]

And that's a gross over-simplification.

I would like to add there are also regulatory biochemical functions, including hormones, that impact the function of many things, often indirectly. Its a big biochemical soup, and requires careful science to figure out the important bits.

If the pathophysiology of ME were simple we would have figured it out a long time ago.

Most initial findings are wrong. Many of those are not published, though I wish there were a repository for unpublished findings as has been discussed. What they do i s set up things for further research. If there is funding and interest we then get follow up studies. Ooops. I used a swear word: funding.

 

[Aileen]

@Valentijn Just to be clear, the mutation is the "A1" column in Table 1 of the paper. Correct??
If so, I have only one homozygous mutation and that is not one considered significant. Most of my results are heterozygous (one copy of each). I believe you have my data, Valentijn.

What is very surprising is that I am homozygous for the normal allele for both significant mutations on the TRPA1 gene that is also mentioned in relation to migraines. Migraines are a huge part of this for me and there is migraine in immediate family member.

Perhaps our results could help identify subgroups? Those who match up with the results vs those who have the opposite for instance?

 

[Valentijn]

@Valentijn Just to be clear, the mutation is the "A1" column in Table 1 of the paper. Correct??

Yes, A1 is the variant they're discussing the prevalence of in patients versus controls. They weren't very clear about it, but based on the SNPs which have minor allele frequencies that aren't near 50%, it's pretty clear that it's A1.

If so, I have only one homozygous mutation and that is not one considered significant. Most of my results are heterozygous (one copy of each). I believe you have my data, Valentijn.

Yup, my little analysis included the data from both yourself and @snowathlete

What is very surprising is that I am homozygous for the normal allele for both significant mutations on the TRPA1 gene that is also mentioned in relation to migraines. Migraines are a huge part of this for me and there is migraine in immediate family member.

Perhaps our results could help identify subgroups? Those who match up with the results vs those who have the opposite for instance?

At this point the SNPs are looking like either a ton of false positives, or just random noise generated from the garbage-in-garbage-out approach of using Fukuda for patient selection. There's really no point to trying to identify subgroups from that.

 

[Valentijn]

Just to mention that in addition to SNPs that affect the function of a protein/enzyme, such as what you're referring to, and which are found in the exon portions of genes (and which code to make the protein), there are other non-coding SNPs on each gene that are potentially just as important and which affect not the function of the resultant protein, but rather how much protein is made and/or at what rate and/or under what circumstances. So it's sort of a quality (how well the protein/enzyme functions) versus quantity (how much and when the protein/enzyme is made) issue. And that's a gross over-simplification.

Yes, but non-coding SNPs always (?) have a pretty small effect size from what I've seen. So while MTHFR C677T +/+ (a missense mutation) might reduce enzyme function by 70%, something like CBS C699T +/+ (a non-coding variant) increases enzyme function by 5%, at most. They also usually have weaker and often contradictory associations with disease, and no reliable direct indication of the enzyme being up- or down-regulated.

Theoretically some non-coding SNPs must be having an impact, but sometimes I seriously wonder, especially after reading a bunch of research where likely false-positives are enthusiastically presented as an amazing discovery. I wouldn't be surprised if 90-95% of the research involving correlations between diseases and non-coding SNPs is completely wrong.

However I do think it's inevitable that there will be a breakthrough eventually, and humanity will find a way to better understand what's happening in those non-coding sections. But for now, it seems like an awful lot of stumbling around in the dark, while grabbing random things and shouting "Look, I found CFS!", etc.

 

[Valentijn]

Personally, with funding for ME/CFS reseach as it is, if I were researching ME/CFS I would probably put my research effort into something other than genetics, unless I had some other reason to suspect a specific mechanism at work that justified looking at the genes involved in that mechanism. But maybe these researchers do?

Genetics research can be relatively simple, so that's part of the attraction. Much of it is basically just the automated processing of a lot of data. It can also be cheap, depending on how much data is being looked at. Whole exome sequencing (the coding bits of every known gene) is down to about $1,100 for the general public - and probably cheaper for researchers having a large amount of samples run.

I think that's why we get a lot of non-genetic researchers playing around with it. Psych papers supposedly linking dozens of non-coding SNPs to personality traits or disorders are especially common. Publishing those sorts of papers is really starting to look like something researchers do when they need to publish to boost the reputation of their institution, and/or keep their employers happy.

As an example, I'm a very sick lawyer who nearly failed high school biology and I'm capable of processing the data in a similar way

What I'd really like to see is full exome analysis of a large group of tightly defined ME/SEID patients. With appropriate correction for false positives! Though given the number of SNPs which would be included, that might need to be a really big group of patients, for statistically relevant results to be possible.

 

[Valentijn]

Interesting. I have been working on the idea that what we most need to ME is a data repository of all results for all research so that positive and negative (including unpublished) replications are all lined up for everyone to see.

I now realise that maybe we already have this repository. It is called Simon and Valentijn, not to mention a number of others. Whenever a question arises it seems that someone has all the old data on tap. But I wonder if it exist in a formal repository of any sort - a sort of Cochrane meta-analysis of research database. If not maybe it should?

The first problem with creating a repository for genetic ME research is that there has been little or no genetic research involving ME patients. It almost all uses Fukuda, which has rather limited relevance. And I have yet to see a Fukuda study where the "risk" SNPs are more frequent in the ME patients on the forum than they are in non-patients. According to those studies, myself and most of the people here have been blessed with a unusually low chance of developing CFS At least they're good for a giggle.

The other problem is that these genetic studies usually have methodological flaws which make them unreliable. Such as the oh-so-frequent failure to correct for the statistical inevitability of false positives when comparing hundreds of measurements at once.

So basically there aren't any reliable positive results - but there aren't any reliable negative results either. There's a bunch of Fukuda crap, and it's mostly methodologically flawed. The closest thing we might have to reliable genetic research data is from the gene expression studies by the Lights following exertion. And that's just expression, not the actual genes, so could be very well be due to completely non-genetic influences.

I'm afraid that my only contribution to a repository regarding genetic research into ME at this point would involve a rude emoticon

 

[nandixon]

Yes, but non-coding SNPs always (?) have a pretty small effect size from what I've seen. So while MTHFR C677T +/+ (a missense mutation) might reduce enzyme function by 70%, something like CBS C699T +/+ (a non-coding variant) increases enzyme function by 5%, at most. They also usually have weaker and often contradictory associations with disease, and no reliable direct indication of the enzyme being up- or down-regulated.

Theoretically some non-coding SNPs must be having an impact, but sometimes I seriously wonder, especially after reading a bunch of research where likely false-positives are enthusiastically presented as an amazing discovery. I wouldn't be surprised if 90-95% of the research involving correlations between diseases and non-coding SNPs is completely wrong.

I'm going to have to disagree with a lot of what you've written.

It's not so important, but just to mention first that CBS C699T is not a non-coding variant. It's a coding SNP in exon 8 of the CBS gene.

I've read enough SNP-disease correlation papers to feel pretty confident to believe that you're fairly incorrect about what you're thinking here. There are plenty of (non-coding) SNPs in the untranslated regions (3'-UTR, 5'-UTR), as well as intronic SNPs (e.g., RNA splicing), of genes that can easily have the same degree of detrimental impact on health as coding SNPs like C677T in the MTHFR gene.

Taking introns as an example, here's an intronic SNP that was shown - in a functional assay - to cause neurofibromatosis:

Functional splicing assay shows a pathogenic intronic mutation in neurofibromatosis type 1 (NF1) due to intronic sequence exonization

And of course, it's well known that UTR SNPs are heavily involved in, e.g., diabetes:

The Role of Single Nucleotide Polymorphisms of Untranslated Regions (Utrs) in Insulin Resistance Pathogenesis in Patients with Type 2 Diabetes

However I do think it's inevitable that there will be a breakthrough eventually, and humanity will find a way to better understand what's happening in those non-coding sections. But for now, it seems like an awful lot of stumbling around in the dark, while grabbing random things and shouting "Look, I found CFS!", etc.

I think you're thinking that we're looking for a SNP(s) as a cause for ME/CFS that is of such great significance that it's going to show up even in a cross-ethnic genome analysis and that it's going to need to be a coding SNP. That might happen, but I doubt it at this point. I think it's much more likely, assuming there is an underlying genetic component to our disease, that it will be a SNP(s) that has an effect not on the function of the resulting protein/enzyme, but rather is somehow susceptible to causing dysregulation of the gene. Just my opinion.

 

[Hutan]

@Valentijn . Results from 23 &Me for my son and I (both ME/CFS) should be back in a month or so. Are you taking more results for aggregation? If so, can you please direct me to the instructions for processing them/providing them to you?

I wonder why researchers don't report global/ethnic population frequencies along with their control data as you have done. Quickly puts the variations into perspective. Ha, maybe that's the reason.

 

[nandixon]

I wonder why researchers don't report global/ethnic population frequencies along with their control data as you have done. Quickly puts the variations into perspective. Ha, maybe that's the reason.

Global allelic frequencies aren't relevant for these types of analyses. Ethnic specific data must be used. See my post here for more information:

http://forums.phoenixrising.me/inde...an-extremely-rare-mutation.37366/#post-593923

 

[Hutan]

@nandixon, I understand that it is the ethnic population frequency that is directly comparable. The global population frequency is quite interesting too though in terms of showing what works for humans in general. E.g. if your patient group has a SNP frequency greater than the ethnic population average but less than the entire human population average, then it's less likely that that SNP is a problem.

My main point was, if that global ethnic data is available, why would you not include it beside your control results?