Supplementary Figure 1: Rescaling of pseudo-time reduces technical impact of sampling density and trajectory building algorithm.

(a) Points distributed along a linear curve in the 200-dimensional space were sampled either uniformly or with bias to increase the density at the trajectory midpoint. Trajectories were built per sampling by Monocle2 algorithm and the resulting pseudo-time scores distributions are shown with color denoting magnitude of sampling bias. (b) Absolute pseudo-time shift elements (n=200 elements) resulting from alignment of the uniformly with the biased-sampled trajectories with (light blue) or without (pink) pseudo-time rescaling. (c) Absolute pseudo-time shift elements (n=200 elements) resulting from alignment of the uniformly sampled trajectory generated by different trajectory building algorithms with (light blue) or without (pink) pseudo-time rescaling. Per boxplot, the center, minimum, maximum, and range denote median, lowest value within 1.5 IQR of lower quartile, highest value within 1.5 IQR of upper quartile, and inter-quartile range, respectively. Abbreviations: DPT: diffusion pseudo time.