Fig. 1
From: Dictionary learning allows model-free pseudotime estimation of transcriptomic data
Spearman correlation for the values of the coefficient vector from DiL for each parameter. Shown are results for five simulation (sub-)pattern/ perturbation combinations in rows for three values of simulated genes exhibiting the pattern (|gsim|) in columns. The dataset labelled ‘Increasing perturbed’ is the one with high-noise-and-zero-counts perturbation, for the other datasets only noise perturbation is conducted. For the dataset with an increasing and a fluctuating pattern, the correlations for each subpattern are shown separately. The respective subpattern is given in the row description in italic letters. The x-axis of each plot shows the number of atoms (m) and the y-axis the atom ID. The maximal correlation increases or remains stable for an increase of the presented values for |gsim| (100,500,1000) with few exceptions. Typically, for each value of m there is one atom that represents the time course