Hi, I am trying to perform an eQTL analysis using Plink. I consider the gene expression of a particular gene as a quantitative trait and run Plink with the following options --linear --genotypic --mperm 100 --mperm-save-all
According to Plink the first line of the assoc.linear.perm file should contain the stats for the original dataset while the following lines contain stats for permuted phenotypes. When I compare the assoc.linear file that contains p-values and statistics and the first line of the perm file I get different statistics and I don't understand why.
Does somebody know why there is a difference between the stats?
Thanks a lot in advance
•
written
by
Rossella •
I would not use PLINK - from what I hear from my colleagues. Take a look at http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1002078 for a recent article on identifying eQTL associations in primary human liver samples.
•
written
by
Larry_Parnell
My reason to use Plink resides in the fact that I want to reproduce results found in a paper and then apply the same technique (which is considered state of the art) to my dataset. The problem is that I cannot easily understand the output given by Plink. I could code everything myself but that would take a lot of time.
Log in to answer this question.
I noticed this as well; the P values from association without permutation are from the Wald test (so say the docs). In principle the corresponding values in the permutation result should be the result of the same point-wise test, but in my results they differ slightly.