Fig. 5: Demonstration of the denoising method.

a Overview image of a mosaic epithelial tissue of stage 7 Drosophila egg chamber containing cells that lack E-cadherin GFP next to cells that are homozygous for E-cadherin GFP. Non-GFP cells are marked by a visible nucleus shown in green. Image is representative of 3 independent experiments. In the super-resolved image epithelial cells homozygous for E-cadherin GFP are shown in magenta next to non-GFP cells shown in green. Borders between cells are illustrated in grayscale. Borders between cells homozygous for E-cadherin-GFP appear as double lines (inset 1) while borders between homozygous E-cadherin-GFP cells and non-expressing cells appear as single lines (inset 2). Borders between non-expressing cells are not visible. b Example cumulative distributions of the times between binding events along with an exponential fit (shown in red) for non-rejected and rejected clusters with low and high numbers of binding events. c–e The fraction of rejected clusters (blue line; total percentage of clusters rejected as the minimum number of binding events required per cluster increases) and cluster rejection rate (orange line: the change in the percentage of clusters that are rejected when the minimum number of binding events is increased by one) for the non-expressing region, the expressing region, and the cell borders respectively. Cluster rejection rates were smoothed using a 5-point moving box median. f Normalized rejection rate of the expressing to the non-expressing region as a function of the minimum number of binding events per cluster. The data were smoothed using a 3-point moving box averaging. The optimal threshold of minimum number of binding events for the denoising method is determined when the normalized rejection rate exceeds 1.