Estimating the mean fluorescence per cell requires several preliminary steps. a) Correcting autofluorescence using the wild-type fluorescence per cell, following the method of de Jong et al. developed for bacteria, can lead to negative estimates in yeast. Correction using the bacterial method is shown in red; correction using spectral unmixing is shown in blue. Data are for cells grown in 2% raffinose and then grown further in 2% raffinose in the plate reader. Raffinose does not induce GAL expression. b) Emissions spectra for autofluorescence and enhanced GFP (EGFP) following excitation at 485 nm. Cells were grown in 2% raffinose and then in 2% galactose for the time indicated. Data for purified EGFP were obtained from the website of the R. Tsien laboratory at U.C. San Diego. c) The OD correction curve: relative cell density obtained from diluting a culture by known factors plotted against the measured OD of the diluted cultures. The blue line shows a regression and the shading gives 95% confidence limits (plus and minus twice the standard deviation of the posterior distribution found using a Gaussian process). d) Estimating r
: the ratio of fluorescence emitted at 585 nm to 525 nm plotted versus OD. The blue line shows a regression and the shading gives 95% confidence limits (found using a Gaussian process). Cells were grown in 2% raffinose and transferred to 1% galactose in the plate reader.