search

kstreet13 / slingshot

Functions for identifying and characterizing continuous developmental trajectories in single-cell data.
259 stars 42 forks source link

pseudotime in two lineage？ #247

Closed handesome closed 3 months ago

handesome commented 3 months ago

Hi！ Thank you for developing the package！when I run the vignette，crv1 <- getCurves(lin1)，the crv1 has lineage1 and lineage2 in crv1@metadata[["curves"]]，so which lineage‘s lambada could be used to calculate the metrics, such as kendall,POS.... Thanks!!!

kstreet13 commented 3 months ago

Hi @handesome,

If I'm understanding correctly, it sounds like you have more than one lineage in your data and you want to compare multiple pseudotime/trajectory inference algorithms? When that's the case, you need to use alternative metrics, because you can't just ignore cells from one of the two lineages. For example, you could calculate a weighted correlation (or other measure) along each lineage and then average them based on the number of cells in each lineage. As I said before, there are multiple papers that discuss such metrics, including the Slingshot paper and the Dynverse paper.

Best, Kelly

handesome commented 3 months ago

Thank you for the quick reply!!