SNPs significant in all Plink tests
0
0
Entering edit mode
3 months ago
gwas_maniac ▴ 20

Hello! I've been trying to perform a GWAS on my dataset using Plink. I've got a good hold on how the basic allelic test and alternate tests work but the majority of my SNPs are significant in both chi squared tests and also in all other model tests. Which makes me a bit nervous on how to interpret and comment on these results. For example SNP1 exhibits significance in basic assoc test, but also is significant when using cochran armitage, and dominant/ recessive tests. How is this interpreted and what's the significance of this result?

Any help is appreciated

SNPs plink gwas statistics • 504 views
ADD COMMENT
0
Entering edit mode

Batch effect in data?

ADD REPLY
0
Entering edit mode

i'm guessing maybe yes.. i have genotyped 7 plates of samples with the same array but 2 more with a different version of the same array.. What would you suggest I do as a "clean up"?

ADD REPLY
1
Entering edit mode

if your cases and controls are mixed between plates - you can use something like COMBAT or any other PCA-based method (I guess you can simply include principal components as covariates)

If cases and controls are not balanced between array versions, then it can be a data autopsy situation.

ADD REPLY
0
Entering edit mode

If cases and controls are not balanced between array versions, then it can be a data autopsy situation.

What do you mean by this?\ Also thank you for the advice

ADD REPLY
0
Entering edit mode

if your cases were analyzed with the platform A and the controls with the platform B, you won't be able to separate biological differences between cases and controls and technical differences between platform A and platform B.

"data autopsy" is the term which is used when a statistician is invited after the experiment was already planned and performed, but the person who performed it lacks knowledge of statistical experimental design. it is often only possible to say why the data is useless.

ADD REPLY
0
Entering edit mode

data autopsy sure sounds like a nightmare.. i will try the methods you suggested and see how it goes... thank you!

ADD REPLY
0
Entering edit mode

once I told a famous professor that mice that they generated with thousands of moneyz are different at the baseline and can't be used as controls and cases and thus the scientific question can't be answered - this one looked like a nightmare XD

ADD REPLY

Login before adding your answer.

Traffic: 2132 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6