Plink Snp- Quantitative Trait Association - P-Value Is 0

Question

[Agatha]

I have run PLINK --assoc for a list of snps and the associated genotypes and qt measurements.

The SNP I have added as control (control= it is known that it is highly associated with the analyzed quantitative trait ) is returned as having a p-value of 0. When doing a separate linear regression analysis in R the p-value is 1.212e-308 .

The genotypes have been used for already published studies.

The p-value is normally very low for that association, but why is it returned as 0 by plink?

Plink returned the 3 expected most significant SNPs (out of 700) as having a p-value of 0.

Answer 1

[Drei]

1.212e-308 is extremely low - I'd imagine PLINK rounds off any P-value low enough to 0. Just how low isn't in the documentation, but if you need to know, I suggest generating some fake data that would give you a range of low p-value and feeding it all to PLINK to see at which point it turns a tiny p-value into 0. Which might be pretty difficult, actually.

[Agatha]

That must be the case ... I have looked in the documentation, forums, examples but I couldn't find anything similar to what I have or any information about the thresholds...