By selecting clones based on criteria such as possessing at least one lineage of interest, for example ILCs, and then clustering these clones according to their lineage outputs, it would be possible to identify progenitor populations with distinct lineage output patterns based on the sharing patterns between ILCs and other lineages.

Is the final stage of ILC specification truly the bifurcation point from the NK lineage?

In addition, it would also be interesting to select clones that retained MPPs and classify them by clustering analysis. Such an approach might provide insight into what types of lineage-output progenitors are present within selfrenewing MPPs.

Filtering based solely on Pgen appears likely to exclude cases where truly unique clones, as well as include cases where non-unique clones are retained. Specifically, in Fig. S3, there are cases with Pgen around 10^2 without high barcode frequency, as well as cases with Pgen greater than 10^4 that lie just above the proportional relationship between BC frequency and Pgen. Because barcode clones detected only once among the n=7 samples contain, on average, at least >85.7% unique clone information, it may be possible to generate a more reliable dataset by combining multiple criteria, such as Pgen, read frequency, and uniqueness across samples.