# Description of WPs

*This is a first draft of spoken text to describe the WPs (Simon's and Ana's part)*

In our data, one sees, in both species, a continuum of cell states: from quiescent stem cells over activated ones, which give rise to transit amplifying cells, and from there to neuroblasts. To maintain stemness, the system has to carefully balance dormancy and activation, and hence, only some of these quiescent cells are ready to be activated. If we were to see into the past of these intermediate cells [point at TAPs], we might discover that they all stem from only a limited part of this reservoir of dormant stem cells [circle part of qNSC]. To understand the heterogeneity in the stem cell pool, to see, which of them can be activated under which circumstances, we need to look into their progenies' past -- and this is what the "time machine" approach allows us to do.

It works as follows: With a doxycyclin pulse, we can induce expression of this fusion protein of polymerase 2 and the bacterial DCM methyltransferase [point at it]. When genes are transcribed by it, a trace of methyl marks is left in the gene body DNA. When we then sequence the cells, several days after the doxy pulse, we get two transcription profiles: the one from the sequenced RNA, giving us the current state [point to TAPs] and the one from the methyl marks, giving us the state during the doxy pulse [point to qNSC]. So, we can learn which sub-compartment in the quiescent stem cell pools had been activated. More generally, we can link stages in feature space and thus get trajectory bundles [point to arrows]. Hence, while standard analysis only reveals the 1D flow along pseudotime, we can now study the heteogeneity orthogonal to the pseudotime direction and observe how it influences lineage progression. We will develop mathematical tools to describe such flow with PDE-like models.

We expect that this heterogeneity is induced by signalling from neighboring cells. Therefore, we are developing methods, where barcode oligos are transferred from a cell to its neighbours -- via propagation of rabies virus and via exchange of exosomes. In subsequent sequencing, we can then regress the transcriptional heterogeneity in, e.g., the stem cells onto the charcteristics of their cellular neighborhoods. We expect that difference in type and state of neighbor cells correlate with differences in cells' transcriptional profile and depth of dormancy.

These techniques, the "time machine" record of past states, and the neighborhood tracing, will give us data to model how state progression is influenced not only by the cell's own state but also by cues from its neighbours.

Complementary to these sequencing approaches, we will use 4D microscopy to track in vivo the progression of cells, in the context of their neighborhood. The transcriptomics data will allow us to construct fluorescent reporter lines to identify precise substates, so that we can map cells that we have observed in 4D microscopy to positions in the scRNA-Seq feature space. In mouse, we use flow cytometry -- but fish are transparent, and so, we can even do this in vivo, tracking cells over days across multiple images. Using this data, we will augment our models to take in the spatial aspect from cellular neighborhoods.