Fig. 2: Biological applications with ECLiPSE in 2D SMLM.

a, Example clusters of the four different tau aggregate species. b, Data exploration using PCA indicates the complexity of the data. c, Classification results obtained (89.8 ± 0.4% accuracy) using the logistic regression classifier on the variable selected tau aggregates data. d, Tau clearance, after removal of Dox, observed over a period of 10 days shows a visual reduction of tau aggregate cluster sizes over time. e, Tau aggregate species prediction on the total dataset demonstrates the rapid decrease in branched fibrils, pre-NFTs and NFTs after Dox removal, whereas linear fibrils show delayed degradation kinetics. FOV, field of view. f, Four representative 2D SMLM images of cells containing two different patient specific TDP-43 strains. g, Representative clusters of the two patient specific TDP-43 strains. h, Data exploration using PCA indicates that a nonnegligible subset of the data clusters are similar between the two patient-specific strains. i, Classification results obtained (89.9 ± 0.6% accuracy) using the partial least squares classifier on the nonvariable selected TDP-43 data. For classification in c and i, only the results obtained by the 100 best models out of 1,000 generated models are shown. For a–e: +Dox control, n = 29 cells; day 1 − Dox, n = 27 cells; day 2 − Dox, n = 27 cells; day − 3 Dox, n = 27 cells; day 4 − Dox, n = 29 cells; day 5 − Dox, n = 27 cells; day 10 − Dox, n = 28 cells; all three biological replicates. The bar plots in e represent mean ± s.d. of the prediction of the 100 best models as shown in c. For f–i strain A: n = 15 cells (three biological replicates) and strain B: n = 19 cells (three biological replicates).