Extended Data Fig. 6: Effect of selecting different source and target states on TPT results.
TPT inference was performed using different source and target states, using the week 15 timepoint of donor 07 from the Kim et al. vaccination time-course dataset. All inferences were performed using the NMF-derived CSR potential as pseudotime input to sciCSR which was invoked using identical and default parameters. The inferred fluxes tend to be higher for transitions involving states which are either the chosen source or target states. Statistical significance was obtained by one-sided comparisons to determine whether the flux values are greater than predictions generated from randomised transitions.
