Bootstrap estimation of confidence intervals on mutation rate ratios

Abstract

The fluctuation test is a useful tool for estimating the mutation rate of cells. However, statistical methods for comparing mutation rate estimates between different strains or conditions have not yet been fully developed. Methods exist for placing confidence intervals on estimates of the number of mutational events in cultures for a given strain and set of conditions, but placing confidence intervals on mutation rate is complicated by differences in the final number of cells in culture between parallel cultures. Additionally, confidence intervals on individual mutation rate estimates are not always the most useful statistical tool when comparing two or more different strains or conditions. We present a bootstrap method for estimating confidence intervals on the quotient of two mutation rates determined from two fluctuation test experiments using experimental and control strains or conditions. We use Monte Carlo simulations to validate this method over a wide range of mutation rates and for empirically measured variation in the estimates of final number of cells in culture. Furthermore, we provide the computational tools to implement the bootstrap method described here on experimental fluctuation test data and to evaluate this method for experimental parameters other than those considered herein.