On determining the power of digital PCR experiments

Abstract

The experimental design that will be carried out to evaluate a nucleic acid quantification hypothesis determines the cost and feasibility of digital polymerase chain reaction (digital PCR) studies. Experiment design involves the calculation of the number of technical measurement replicates and the determination of the characteristics of those replicates, and this in accordance with the capabilities of the available digital PCR platform. Available digital PCR power analyses suffer from one or more of the following limitations: narrow scope, unrealistic assumptions, no sufficient detail for replication, lack of source code and user-friendly software. Here, we discuss the nature of six parameters that affect the statistical power, i.e., desired effect size, total number of partitions, fraction of positive partitions, number of replicate measurements, between-replicate variance, and significance level. We also show to what extent these parameters affect power, and argue that careful design of experiments is needed to achieve the desired power. A web tool, dPowerCalcR, that allows interactive calculation of statistical power and optimization of the experimental design is available.