is it possible to choose one cell as the starting point for Slingshot

[liuweihanty]

Hi there! I'm wondering is it possible to choose one single cell as the starting point to construct lineage?(I only want to construct one lineage) From my understanding it is only possible with clusters? Thank you very much!

[kstreet13]

Yes, it is possible! I've actually been thinking about this recently and I think it could be quite useful! Here's how I would do it:

• Change the cluster label of the starting cell to something unique, essentially turning it into its own cluster.
• Set that cluster as the starting cluster.
• Use dist.method = "mnn" because the default distance metric tries to use cluster shapes, but is less accurate in the presence of small clusters.

Let me know if this doesn't work or if there are any additional considerations I missed. Best, Kelly

[liuweihanty]

Thank you @kstreet13 ! that is cool! Just make sure I understood you correctly, I have a small population(~1000 cells), and I just want to infer one lineage on this population. Therefore, would it make sense if I label this starting individual cell as cluster 1, and the rest 999 cells as cluster 2? (Because this populations of cells are relatively homogeneous and I don't want to further cluster the rest 999 cells before running slingshot.). Thanks!

[kstreet13]

Yeah, I think that makes total sense! In cases like this, Slingshot is nearly identical to a standard principal curve, so it sounds like this would basically just be a way to ensure that the curve has the correct orientation.

[liuweihanty]

Thanks very much @kstreet13 , indeed, I just wanna ensure the directionality by choosing the starting cell.