Table 3 Comparison of different pooling schemes and total cost using model data. Several pooling designs that can achieve power at least 0.8 while controlling type I error rate at 0.01 for an effect size of 1.0 are shown. Assuming a microarray chip costs $1000 and a subject costs $300, the total cost for each design is also computed and the optimal design with the minimal total cost is underlined. A function written in R (a free statistical software downloadable at http://www.r-project.org) to perform the above search automatically is attached as additional file.

From: Statistical implications of pooling RNA samples for microarray experiments