We benchmarked Frapbot using a previously published FRAP dataset obtained with inducible human embryonal
kidney (HEK293) cells that expressed a monomeric green fluorescent protein (AcGFP1) in the mitochondrial
matrix (Dieteren et al., PNAS, 2011). In the original study, the average trace of multiple FRAP measurements (n=187) was manually fitted
with a single component exponential model by Origin Pro software, delivering a τmono value of 0.600 s,
corresponding to a t1/2 of 0.416 s. Individual analysis of the identical FRAP traces by Frapbot (Fig. 3, A) and
subsequent averaging of the obtained t1/2 values, delivered a slightly higher mean halftime (t1/2) of 0.463 s (Fig.
3, B). This difference might be explained by the larger noise of the individual traces compared to the averaged
FRAP curve, that influences the quality of the fitting. Frapbot analysis suggests that the t1/2 values of the dataset
were not Gaussian distributed. If this is the case, we recommend the use of the median instead of the average for
the analysis of similar future FRAP data which provides more robustness against outliers. Consequently, the
calculation of the median was implemented in Frapbot.