unifiedWMWqPCR: analyzing RT-qPCR data in R with the unified Wilcoxon–Mann–Whitney test

Most conventional statistical tests for analyzing RT-qPCR data require normalization before differential expression can be assessed. This normalization can have a substantial effect on the interpretation and validity of the statistical test, but this effect is often ignored. Therefore the uWMW test, as proposed in [1], extends the Wilcoxon–Mann–Whitney test so that the normalization is incorporated in the testing procedure. Both the effect size and the normalization have an interpretation in terms of the probability P (Y 4 Y ′) := P (Y < Y ′) + 0.5P (Y = Y ′), where Y and Y ′ denote independent responses (here quantification cycles). We employ the same notation as in [1]. Let the random variable Yijk denote the quantification cycle Cq associated with feature i ∈ {1, . . . ,m + h} (a feature can for example be a gene or a microRNA) of sample j ∈ {1, . . . , nk} (a sample can for example be a tissue or a patient) in