Correcting p-values to account for multiple hypotheses

In bioinformatics, particularly in genomics projects, we often perform statistical tests thousands of times in an analysis. But this can be a source of significant error in our results. Consider a gene expression experiment that has small numbers of measurements per treatment (often only three) but has tens of thousands of genes. A user doing a statistical test at p <= 0.05 will reject the null hypothesis incorrectly five percent of the time. Correcting for performing multiple hypotheses allows us to reduce the error rate from such analyses. We will look at a simple-to-apply method for making such a correction.

Getting ready

All of the functions we need are base R and...