Power calculation for a collapsing rare variant analysis
0
0
Entering edit mode
3 months ago
tacrolimus ▴ 100

Dear Biostars,

I have 250 cases with a disease assumed to be autosomal dominant and 19000 controls. I have performed a exon wide, collapsing gene based rare variant (MAF<0.001, CADD>20) analysis with SAIGE0GENE which did not yield any hits.

The disease in question is assumed to be fully penetrant.

I would like to know a)am I underpowered to detect/discover even a single associated gene (I assume yes but would like to qualify that) b) what kind of numbers would I need to reach 0.8 power.

I have looked at genpwr in R (https://cran.r-project.org/web/packages/genpwr/index.html) as well as the PAGEANT web interface (https://andrewhaoyu.shinyapps.io/PAGEANT/) but have found some of the terms confusing and was wondering if anyone had any experience with them specifically:

In genpwr when using the power calculator I am asked to give an Odds Ratio. I know what that is in theory, but given it's an AD disease what would one put down as the OR? Equally would you put the alpha as 0.05 or the genome wide one (assuming Bonferroni 0.05/number of genes), at present my code for this looks like:

pw <- genpwr.calc(calc = "power", model = "logistic", ge.interaction = NULL,
N=19701, Case.Rate=0.01, k=NULL,
MAF=seq(0.0001, 0.001), OR=c(10),Alpha=0.05,


For PAGEANT under the "case-control" tab I have opted for the "single gene" panel with an alpha set to 0.05 and the "EV:variance explained" set to 80% as I assume most of the disease comes from the genetic rather than environmental make-up - does that seem reasonable or should I be using the whole genome tab

Your thoughts would be greatly welcome!

All the best