A statistical method for comparing viral growth curves

Viral replication is often analyzed by growth curves, in which viral multiplication in the presence of host cells is measured as a function of time. Comparing growth curves is one of the most sensitive ways of comparing viral growth under different conditions or for comparing replication of different viral mutants. However, such experiments are rarely analyzed in a statistically rigorous fashion. Here a statistical method is described for comparing curves, using replication of HIV in the presence of an integrase inhibitor as an example. A complication in the analysis arises due to the fact that sequential measurements of virus accumulation are not independent, which constrains the choice of statistical method. In the recommended approach, the values for virus accumulation over time are fitted to an exponential equation, then the means of the extracted growth rates compared using a nonparametric test, either the Mann-Whitney U-test for two samples or the Kruskal-Wallis test for multiple samples. A web-based tutorial for implementing this method is available at http://microb230.med.upenn.edu/tutorials/wangTutorial.