The full spectrum of genetic architectures represented by gene expression traits provides a unique opportunity to empirically assess the power of different designs for association studies. We have shown here, for example, that reduction in gene expression measurement noise by PC analysis can markedly improve the ability to identify novel eQTLs.
