Power and sample size calculation for microarray studies (Fortran).

The Fortran program is to calculate the sample size of a microarray project to discover the prognostic genes that are associated with a binary endpoint while controlling the familywise error rate (FWER). Since the gene discovery by controlling the FWER is sensitive to the dependency among the genes, the authors propose a two-stage design for microarray projects. At stage 1, we conduct a pilot study to generate a pilot data. At stage 2, we generate a large number of simulated microarray data with a dependency structure among the genes the same as that of the pilot data, and estimate the sample size of the whole project for specified number of prognostic genes, their effect sizes, the FWER level and the overall power.