Fig. 3

Population analysis of total screening data. Even with added genes, the two background strains perform differently as determined with antibody titer (a) and optical density (b), forming two populations of measurement values. Δopi1 (dashed red line) had generally higher antibody concentrations (a) and lower OD₆₀₀-values (b) than the wild-type (black solid line) at all temperatures tested. Distribution of the specific product yields calculated separately for each sample and replicate are shown in (c). The distribution of values is shown as density plots, where the area under the line is equal to 1 (total probability of a measurement value being under the line) and the y-axis is proportional to relative frequency. The range of the measured values is given in the x-axis. Each condition included the analysis of 113 different strains and the number of measurement points in each curve is between 798 and 824 in (a) and 399–412 in (b) and (c). The mean values for each strain were normalized to the respective background strain and plotted together to represent all the measured values (d). A linear regression plane (dashed lines) was fitted to the data points to model the dependency of antibody titer on combined effects of OD₆₀₀ and normalized specific product yield. Multiple R-squared was equal to 0.8896 indicating that a linear dependency was found. The estimated coefficient of fold-change was higher than that for the normalized OD₆₀₀, meaning that a change in specific product yield had a higher impact on the antibody titer values. Colors are added only to aid the visual clarity in (d)