Comparison of estimated to true subpopulation sizes in simulated bimodal populations by using different separation methods.

1True subpopulations were simulated using the R function rnorm(…) with a standard deviation of 37.7, a mean value of 127.3, and the number of observations corresponding to the subpopulation percentage to be tested from a total number of 2000 observations. Mean and standard deviation used for the simulations represent population parameters as obtained from fluorescence microscopy analysis of batch grown P. knackmussii B13 Pint-egfp in 3CBA (see Table S3).
2Estimated subpopulation sizes (mean ± SD; 3 independent repetitions) were determined applying the PS methods on simulated bimodal populations using the R function findsub(…) (Protocol S2). A bimodal population was simulated by mixing two simulated populations, a real subpopulation1 and a second subpopulation. The second subpopulation was created using the R function rnorm(…) with a standard deviation of 3.9, a mean value of 63.0, and the number of observations depending on the sample size of real subpopulation1 to give a total of 2000 observations. Mean and standard deviation used for the simulations represent population parameters as obtained from fluorescence microscopy analysis of batch grown P. knackmussii B13 Pint-egfp in 3CBA (see Table S3).