Interpreting the output of score test using GENESIS
1
0
Entering edit mode
@lassanaasamarakoon-22071
Last seen 20 months ago

Hi all, I have output from an association test, using a Score test, from GENESIS software. My output has the following variables: Score Var Score.Stat Score.Val

According the GENESIS manual, I was expecting “Score.SE” and not “Var” variable. I want to confirm if the “Var” is Variance? If so, can I use √Var/√n to calculate the SE?

genesis output score • 351 views
2
Entering edit mode
@stephanie-m-gogarten-5121
Last seen 7 weeks ago
University of Washington

You must be using an old version of GENESIS, but looking at the manual for the current version. Var is indeed the variance, and Score.SE = sqrt(Var) (no factor of n).

0
Entering edit mode

Thanks so much, Stephanie!

0
Entering edit mode

Hi, Is there anyway to convert the Beta estimate to a z-score here (equivalent to Score.Stat)?

0
Entering edit mode

I'm not entirely sure what you're asking here, since the Score test doesn't give a beta estimate. Is this what you're looking for? https://support.bioconductor.org/p/119621/

0
Entering edit mode

Hi Stephanie, I'm so sorry for the confusion. I want to do the opposite of the above link you sent. 1)I want to convert the beta estimates to get a z score?

2) I want to know what is the z score variable of the assoc test output, is it "Score" or "Score.Stat"?

Thanks so much for your help.

1
Entering edit mode

1) The score test does not return a beta estimate, but you can get a good approximation of beta by taking Score/Score.SE^2); or, in your case, by taking Score/Var

2) Score.Stat is the test statistic, which is what you are thinking of when you say "z score variable". The current version of GENESIS returns a Score.Stat that is approximately Normally distributed (i.e. a z score); however, since you are using an old version of GENESIS, I can't remember if it returns that, or the square of that, which would follow a chi-square distribution with 1 d.f. To double check, the "z score" test statistic can be calculated as Score/Score.SE (or in your case Score/sqrt(Var)).