Fig. 6: Multiple logistic regression of competing wing discs reveals tissue parameters influencing loser cell death.
a–f Predicted effects plots of logistic regression analysis of individual cells in competing wing discs shown in Supplementary Data 2. In these, the probability of cell death is plotted against the pouch volume (a), the total number of RpS3+/− loser cells in the sample (b), the number of Dcp-1-positive wildtype winner cells in the sample (c), the Z-level of the cell (d), the distance of the cell from the centre of the pouch (e), and the distance of the cell from the loser patch perimeter (f). The plots in (a–f) show the expected probability of apoptosis upon changing a given predictor variable – the distance of the cell from the patch perimeter, whilst all other predictors are set to their mean value. Pale blue region denotes the 95% confidence interval, and black tick marks on the x-axis denote observed values in the dataset. The red dashed line in (f) is at 8 µm, which corresponds roughly to 2 cell diameters from the patch border and denoting the patch border/patch centre cut-off used in other analyses in this study. g, h Comparison of replicates/experiments used in the logistic regression analysis shown in Supplementary Data 3. g The percent loser patch coverage of the wing disc pouch region for each wing disc is shown on the y-axis, while the x-axis shows the replicate/experiment to which they belong, labelled by the date. p-values and g-values reflect two-sided student’s t-tests with FDR correction and Hedges’ g effect size metrics relative to the basemean. Measure of center and error bars are shown as mean and 95% CI, respectively. Biologically independent samples are, in order from left to right, n = 8, 10, 10, 13, 10, 8, 7, 9, 15, 15, 9, 12, 16, 16, 12, 8, 5. h The percentage of Dcp-1-positive loser cells is shown for each replicate/experiment. Statistics reflect either a two-sided Mann–Whitney U-test with FDR correction along with Cliff’s δ effect size metric, relative to the basemean (no border) or the logistic regression analysis along with odds ratio effect size metrics (OR) shown in Supplementary Data 2 (red border). Measure of center and error bars are shown as mean and 95% CI, respectively. Biologically independent samples are, in order from left to right, n = 8, 10, 10, 13, 10, 8, 7, 9, 15, 15, 9, 12, 16, 16, 12, 8, 5. Datasets that were outliers by Mann–Whitney U-test are shown in blue, and red asterisks reflect datasets that had significant p-values by logistic regression but did not meet the effect size cut-off. Source data are provided as a Source Data file. OR odds ratio.
