Why are my p-values deflated (GWAS)?
1
0
Entering edit mode
solanum • 0
@solanum-22356
Last seen 20 months ago

Hi

I performed genome wide assoctiations (linear mixed model, using GENESIS) with about 640'000 SNPs and transformed phenotypic traits of 160 individuals. After correcting for population structure and kinship, all resulting Q-Q plots looked more or less like this:

Q-Q plot example

Does anybody have an idea what could possibly cause deflated p-value distributions in GWAS? Changing the number of PCs and adding different random effects did not substantially improve the model.

Many thanks in advance

gwas genesis p-value q-q-plot • 1.2k views
ADD COMMENT
1
Entering edit mode

Hello,

Have you performed any filtering on your 640,000 SNPs? You probably want to filter on quality, but also minor allele count (MAC). Single variant association tests for SNPs with very low MAC (e.g. < 10) generally will not perform well. Removing very low MAC SNPs may help your deflation issue.

Also, is your phenotype a quantitative or binary outcome? If it is binary (e.g. case/control), you may want to try the logistic mixed model by setting the family parameter = 'binomial'.

-Matt

ADD REPLY
0
Entering edit mode
Anna • 0
@Anna-24693
Last seen 8 months ago

It is also notable that your sample size is extremely small, which wouldn't lead to this qq-plot necessarily but is a problem more generally. The negative trend signifies something wrong with the data.

ADD COMMENT

Login before adding your answer.

Traffic: 328 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