Hi,

I have a dataset where each experimental condition has 3 biological replicates and each biological replicate has 2-3 technical replicates. When I use lmFit(), I can account for correlations between technical replicates using the correlation parameter. However, camera() does not seem to have such a parameter. Do I need to somehow aggregate technical replicates prior to running it?

Thanks.

Yury

I haven't had time to implement the correlation argument for camera(). In the meantime, you could either average the technical replicates (using avearrays) or else just treat the technical replicates as if they were biological replicates. The latter sounds bad, but actually will usually be fine for a competitive test like camera. camera() compares genes, so the genewise t-tests don't need to follow a t-distribution under the null hypothesis.