betas for binary outcome
1
0
Entering edit mode
@stephanie-m-gogarten-5121
Last seen 22 days ago
University of Washington

I am currently working on a project with a binary outcome, and need to report beta coefficients (representing the log-odds increase per copy of the minor allele), confidence intervals, and odds ratios for each variant. I am using a SeqVarGDSClass object with a SeqVarIterator, and have run fitNullModel and assocTestSingle with no problems. Since my outcome is binary and I am using the “family=binomial” option in fitNullModel, the association test produces Score test statistics only.

My question is – do you know if it is possible to get beta coefficients and standard errors for an additive model with a binary outcome using GENESIS?

genesis • 415 views
0
Entering edit mode
@stephanie-m-gogarten-5121
Last seen 22 days ago
University of Washington

You have two options for calculating betas for binary outcomes.

Option 1) refit the null model including the genotypes for the variant of interest as a covariate. This is a good option if you only need betas for a small number of variants. You can use SeqVarTools::altDosage to get the alternate allele dosage for a given variant or set of variants.

Option 2) approximate the betas based on the Score test, using the method in Zhou, Shi, and Whittemore 2011 (https://www.ncbi.nlm.nih.gov/pubmed/21922536).

assoc <- assocTestSingle(iterator, nullmod)
assoc$beta <- assoc$Score/(assoc$Score.SE)^2 assoc$beta.SE <- 1/assoc\$Score.SE