Power Calculations: Snp X Environment

Question

[Bdgs]

I would need to calculate statistical power for models where we have genetic marker (SNP) interacting with environment (continuous variable), and the outcome is a continuous variable as well.

I haven't been able to find any statistical package that would perform calculations like this (there are lots that allow power calculations for simple SNP association, or SNP x environment when the outcome is binary), and am under the impression that this kind of calculation is not trivial (i.e. might require simulations of some kind).

To sum my question up: How should I perform this type of power calculation: SNP x Environment (continuous) = Outcome (continuous)

All help, suggestions and links would be warmly welcomed!

Answer 1

[Stephen]

I'm fairly certain Quanto will do this for you.

[Larry_Parnell]

Yes, a +1 from our group where Quanto is our choice. You must calculate one SNP at a time; batch mode is not possible.

Answer 2

[Larry_Parnell]

You should run the power calculations both ways - with the enviro term as continuous but also categorized (typically around the mean) just to know what the result is. We often look at gene-environment interactions under both situations, say with percent energy in the diet from saturated or mono-unsaturated fat. Thus, we do both the continuous and categorical power calcs up front.

[Khader Shameer]

Nice tip, Larry !