GG rs9472817 annuls APOE epsilon 4 risk? Wo Hoo!

[Hubbs]

If say it aloud could be helpful to others here, then yes

[J11]

Yeah, you have a big smile for a good reason!

Two family members also have your same genotypes for the below SNPs.

rs9381468 (GG)
rs9296504 (CC)
rs9381469. (AA)

They also have:
SLC25A27, LOC100131283 46640992 rs9472817 C or G GG

Looks like someone just got their Get out of Jail card! Woo Hoo!
It's going to be tough for quite a few of us waiting for the replication for this.

Since we have an ante on the table, perhaps my idea of creating our own SNP database makes some sense.

This would be especially important for the 44s.
It is very unclear to me whether this would apply to 44s, the present study would have only had a few
44s with GG genotype to work with (200*0.02*0.15 <1 ?).

[Matisse]

J11

Are the snp results above the reduced risk ones -- GG, CC, AA?

[J11]

Yes, that is the way it worked for us.

If you look at the GMAF for 1468 you will see the G allele is equal to about 45%. {The GMAF for 817 is 38% G.}
There is some wiggle room there where 1468 is not always going to fit 817, though it is quite close.
I also looked to see how close 1468 fit the other 2, and there was difference between the others.
The guess would be that the closer you fit all three of the above SNPs homozygositely, then the higher
the probability of a GG homozygote genotype for 817.

I am not totally sure how you would put all three of the SNP genotypes together to get your best prediction, though
simply using the highest RSquared 1468 and relating its G allele --> the G allele of 817 should give you very good predictive ability.

I would also like to know how you might take the RSquared number and create a probabilistic bound for predicting the 817 genotype. All the same the 0.96 RSquared for 1468 gives you a fairly good feeling.

[J11]

The big question I want to put out there for consideration by the people is:
Should such a highly speculative SNP such as rs9472817 be on snpedia?
User SNPedia was quite unamused by including an unreplicated result on snpedia.
It seemed that this user was contemplating editing out the SNP.

And sure, the point is well taken. Look on alzgene: nearly every single one of those 2973 polymorphisms were wrong!
http://www.alzgene.org/

After 20 years of almost 100% failure, a list of very highly statistically valid AD single variants have finally been found.
Now, we are upping the ante and going for the duo combos.
One should not be totally surprised that the past is prologue and many years of 100% failure will ensue: including rs9472817.
The present study was using a small sample, perhaps a little before the fact data massage was involved.

What do the people think?
Should SNPedia take the high road and remove the SNP?
Would it be worthwhile for someone to add in the rs9381468 so that people on the v4 platform would be informed?

[J11]

Hmm, why reinvent the wheel?

openSNP has almost 5000 members who have deposited there genechip files.
I counted about 150 GG rs9472817 genotypes, and there would probably be quite a few more 1468 as well.
https://opensnp.org/snps/rs9472817#users
https://opensnp.org/snps/rs9381468#users

We could then cross reference there APOE genotype.
https://opensnp.org/snps/rs429358#users
https://opensnp.org/snps/rs7412#users

The only hold back would be that probably most of them are not of dementia age.
This is the sort of database that would be very helpful for us.
It is too bad that it does not seem to be organized in such a way that the information we were interested could be easily extracted.

[Stavia]

Too many big words and too many numbers
J11 I'm so glad you are onto this. My brain is too small

[J11]

Yeah, too many big words might be about right, though too many big numbers?
Big numbers are a good thing!

[J11]

openSNP is getting close to what we want:

https://opensnp.org/search?utf8=%E2%9C% ... phenotypes

What we really really want to see, though, is epsilon 4s with GG genotype and their dementia status during aging.
Obviously epsilon 44s with GG genotype and dementia status would be even more informative.
We would be expecting that these people should have a low dementia burden.

epsilon 4s or epsilon 44s who had C- genotype who did get dementia, isn't really that informative.
While those with these genotypes who did not get AD would provide contrary evidence.

[Hubbs]

@J11, thanks! Wow I finally get a good news for a break... from a month of depressing failures at keeping my 1h pp BG below 140.