In cases the same cluster is estimated as initial and terminal states

[st-tky]

... Hi Cellrank team,

Thank you for developing this wonderful tool! I have a general question about initial and terminal state estimation using GPCCA.

In short: I am wondering if it is acceptable that the same cluster is estimated as initial and terminal states. If it is acceptable, that would be OK for me. If not acceptable from the theoretical point of view, I would appreciate any suggestions to workaround.

The following is a supplemental explanation: I am following the instruction of "Computing Initial and Terminal States" on the website now. In my datasets, the same cluster is estimated as initial and terminal states. Because of the overlap, I need to set an option as follows: g.predict_initial_states(allow_overlap=True) I understand that is why I get the same cluster, but I have to do so; otherwise, the error will come up due to the overlap. In my case, it may be possible that stem cells (initial state) can differentiate into other cell types (terminal states) and also self-renew (possibly terminal state, too). Also, it may be possible that de-differentiation occurs or that other stem cells can transform into the other type of stem cells. Because I am trying to find a new stem cell population that is currently not fully appreciated, it is very difficult to judge if this result is not correct or biologically acceptable.

I would appreciate it if you could give me any suggestions or explanations about it.