(A) Distribution of absolute expression differences observed between EMS and single-site mutants (bars). To assess the statistical significance of these expression differences, we estimated the magnitude of expression differences expected to arise by chance between genetically identical strains grown at different positions of a 96-well plate (red line). This null distribution was obtained from the differences in expression measured for 10,440 pairs of the un-mutagenized progenitor strain grown at different well positions in four replicate populations. We next randomly permuted 105 times the expression values between (i) each pair of EMS and single-site mutants and (ii) random pairs of the progenitor strain to calculate the one-sided p-value for each pair of mutants (i.e. the proportion of randomized expression differences greater than the observed expression difference). After Benjamini-Hochberg correction for multiple testing, we found that the expression difference between the single-site mutant and the EMS mutant carrying the same mutation was statistically significant (adjusted p-value < 0.05) for 14 out of the 40 pairs of mutants (35%, red and blue bars), but highly significant (adjusted p-value < 0.01) for only one pair (2.5%, red bar). Because mutant strains were exposed to the same micro-environmental and technical variation as the control samples used to establish the null distribution, these sources of variation are unlikely to explain the significant differences of expression observed between EMS and single-site mutants. Panels (B–F) test three other hypotheses to explain expression differences observed between single-site and EMS mutants. (B) Hypothesis 1: expression differences between EMS and single-site mutants are explained by differences in expression noise (i.e. the variability of expression observed among genetically identical cells grown in the same environment) among mutants. To test this hypothesis, we compared the expression noise measured by flow cytometry for each EMS mutant (x-axis) to the absolute difference of median expression levels between this EMS mutant and the corresponding single-site mutant (y-axis). We observed no significant correlation between the two parameters (r = 0.06, p-value = 0.71), indicating that expression noise is unlikely to explain expression differences between EMS and single-site mutants. Expression noise was calculated for each sample as the standard deviation of expression among cells divided by the median expression and it is reported as the average value among four replicate populations relative to the expression noise of the wild-type progenitor strain. Dot colors: p-values as shown in panel A. Dot shapes: circles represent mutations identified by BSA-seq; triangles represent mutations identified by sequencing candidate genes. Error bars: 95% confidence intervals calculated from four replicate populations. (C–D) Hypothesis 2: expression differences between EMS and single-site mutants are explained by additional mutations present in the EMS mutants. (C) Testing effects of additional mutations associated with fluorescence: boxplot comparing the magnitude of expression differences when only one mutation was associated with fluorescence and when more than one mutation was associated with fluorescence in BSA-Seq experiments. The fact that no statistical difference was observed between the two classes (Mann-Whitney-Wilcoxon test, p = 0.192) suggests that expression differences between EMS and single-site mutants were not likely to be caused by additional mutations associated with fluorescence in the BSA-Seq data. (D) Testing effects of additional mutations with statistical support for an association with fluorescence below the significance threshold. Expression difference between EMS and single-site mutants (x-axis) was compared to the highest G-value that was below our significance threshold for considering a mutation to be associated with fluorescence in the BSA-Seq data from each mutant (y-axis). A significant correlation was observed between the two parameters (Pearson’s r = 0.48; p = 0.02), suggesting that some mutations with associations below our detection threshold in the BSA-Seq experiments might contribute to expression differences observed between EMS and single-site mutants. Dots represent individual pairs of EMS and single-site mutants sharing the same mutation (with random jitter). The red line represents the linear regression of the y-axis parameter on the x-axis parameter. (E–F) Hypothesis 3: expression differences between EMS and single-site mutants are explained by secondary mutation(s) or epigenetic changes that occurred during construction of single-site mutants. To test this hypothesis, we isolated two independent clones for 26 single-site mutants after transformation of the progenitor strain and measured the expression difference between the two clones. (E) A positive correlation was observed between the expression difference between EMS and single-site mutants (x-axis) and the expression difference between the two independent clones for each single-site mutant (y-axis). This positive correlation indicates that mutations with larger expression differences between the single-site and EMS mutants tended to also show larger expression differences between independent transformants. Dot colors: p-values as shown in panel A. (F) Boxplot also showed that the average magnitude of expression differences between independent clones was higher for single site mutants with a statistically significant expression difference between the single-site and EMS mutant sharing the same mutation (Mann-Whitney-Wilcoxon test, p = 0.008). Results from E and F suggest that secondary mutation(s) and/or epigenetic changes that unintentionally occurred in some of the single-site mutant clones likely contributed to expression differences between some EMS and single-site mutants. It is important to emphasize, however, that these expression differences were small in magnitude and that overall the expression level of single-site mutants was strongly correlated with the expression level of EMS mutants (Figure 3).