Efficient, robust, and versatile fluctuation data analysis using MLE MUtation Rate calculator (mlemur)

Abstract Almost 80 years since the publication of the seminal work by Salvador Luria and Max Delbrück, the fluctuation assay has remained an important tool for analysing the levels of mutagenesis in microbial populations. The mutant counts originating from some average number of mutations are usually assumed to obey the Luria–Delbrück distribution. While several tools for estimating mutation rates are available, they sometimes lack accuracy or versatility under non-standard conditions. In this work we developed extensions to the Luria– Delbrück protocol to account for phenotypic lag and cellular death with either perfect or partial plating. Hence, our novel MLE MUtation Rate calculator, or mlemur, is the first tool that provides a user-friendly graphical interface allowing the researchers to model their data with consideration for partial plating, differential growth of mutants and non-mutants, phenotypic lag, cellular death, variability of the final number of cells, post-exponential-phase mutations, and the size of inoculum. Additionally, mlemur allows the users to incorporate most of these special conditions at the same time to obtain highly accurate estimates of mutation rates and p values, confidence intervals for an arbitrary function of data (such as fold) and perform power analysis and sample size determination for the likelihood ratio test. We assess the accuracy of point and interval estimates produced by mlemur against historical and simulated fluctuation experiments. We believe both mlemur and the analyses in this work might be of great help when evaluating fluctuation experiments and increase the awareness of the limitations of the widely-used Lea–Coulson formulation of the Luria-Delbrück distribution in the more realistic biological contexts. Author Summary Despite many recent developments in the department of fluctuation data analysis, geneticists are often limited to either a very strict classical Luria–Delbrück protocol, or possibly a modified one with relaxation of just one assumption (e.g., differential fitness of mutants and wild-type cells). In some cases, such as partial plating, researchers use historical methods whose accuracy strictly depends on the conditions of the experiment and thus are not optimal in a wide range of parameters. A novel tool for fluctuation data analysis, mlemur (MLE MUtation Rate calculator), alleviates these problems and provides new extensions that allow to account for phenotypic lag and cellular death. Additionally, we find that the failure to properly account for these additional parameters might lead to inaccurate point and interval estimates of the mutation rate. Our results underline the importance of careful examination of the fluctuation assay to ensure that the chosen statistical model reflects the biological and technical conditions of the experiment.