Fig. 1. Illustrating the first two steps of the condiments workflow with several scenarios.

We have four different toy datasets, one per column, with their reduced-dimensional representation displayed in the first row. Each dataset contains 1, 000 cells equally distributed between two conditions: control in blue and knock-out (KO) in orange. They represent possible outcomes from the first two steps of the condiments workflow. The first dataset has no differences between the two conditions. In the second dataset, the KO accelerates the differentiation process: we therefore have differential progression, but neither differential topology nor differential fate selection. In the third dataset, the KO blocks differentiation in the bottom lineage: we therefore have differential fate selection and differential progression, but no differential topology. Finally, in the fourth dataset, the KO modifies the top lineage: we therefore have differential topology. The last row of the table indicates case studies corresponding to each of the scenarios.