Hi @
Valentijn
But what if the SNP´s that @
Sea found connected to mitochondrial impairments are the same for other PWME
? I have gene test results from about 15 people who I know pretty well with their sick stories and it would be interesting to check the SNP´s also for them.
I've been comparing the rare and very rare results from the full 23andMe files of ME patients (43 so far, + 2 more definitely coming), and looking for similarities. I need to get a more powerful PC to process this much data in more useful ways, and to allow for incorporating data from various databases (minor allele frequencies, gene names/locations, pathogenic flags, missense/stop-gain mutation flags, maybe even gene keywords, etc).
One thing I want to do is automatically compare allele frequency for every SNP I have data for, with what is expected and what the "controls" have. I was (barely) able to do this when I had a much smaller sample several months ago, and eventually came up with a list of SNPs sorted based on which SNPs on which we differ the most from controls. As an example, at the top of the list is rs4557033 where we had 17 more instances of the "A" allele compared to the controls. that's quite a lot but the A allele has 38.1% prevalence in the general population. So we we have a MAF of 58.1%, which is a lot higher than the controls at 27.4%, but isn't a huge difference when compared to 38.1%, if you consider that 960,000 SNPs were processed in that way and some random big differences are likely to turn up.
Hence I need to get a statistics program talking nicely to a proper database program, so that statistical significance can be accurately determined, and give us a good idea of what is background noise and what might actually be relevant. Additionally, it is obvious that there's no single SNP in the 23andMe data which is effectively flipping a switch and causing ME/CFS. Hovever there might be a variety of SNPs on the same gene or on related genes which create the same susceptibility - but that involves more complexity to assess, and a need for the appropriate programs (database) and processing power.
Pending the ability to easily automate this sort of processing, I'm compiling lists of shared rare SNPs of ME patients, then using that to expand to looking at the entire gene involved with each shared rare SNP, and genes which share a similar function. Currently I'm doing this by starting with SNPs which have a 1% prevalence or less, with the list for each patient generated by
http://sourceforge.net/projects/analyzemygenes/ and then the list for every patient added to the same excel sheet, where I can sort and tally the number of duplicate rare results. At the top of the list when I had data for 40 patients was rs11608105, which has an expected MAF of 1%, but we have at 25% (8 heterozygous patients, and 1 homozygous). Interesting! But the expected MAF in europeans is actually 7.5%, so not quite as impressive. But if I look at the 31 patients I currently have ethnically roughly-matched controls set up for, we have 22.6% MAF and those controls just have 6.5%. So definitely worth looking into the associated gene (CADM1) and such a bit further.
Anyhow, I am looking for associations in a variety of ways, but right now it's slow and tedious and requires my brain to be working fairly well. Except I'm getting fevers and intense hypotension for nearly the entirety of every day while on a couple antibiotics for Lyme, so I can't really handle it currently. But medical bills have been almost entirely paid and reimbursed now by my parents, so hopefully we can get a proper computer in the next week or so! The fiance actually brought it up last night, so fingers are crossed
