statOmics/tradeSeq

Question about lineage assignment

KemperNiklas opened this issue · 2 comments

KemperNiklas commented

Hi,

I just wanted to ask if there is a reason why you first draw for every cell a lineage (based on the weights) and not just directly work with the probabilities, i.e. why is Z a binary matrix and not a matrix of probabilities.
Additionally, is the Single Cell Experiment output preferred in order to use a common widely used object? I am asking because in the code it looks like new prediction seem quite cumbersome with the see object (compared to a list of GAMs as output).

Thanks in advance!

koenvandenberge commented

Hi @KemperNiklas

It is theoretically certainly possible to use the probabilities as such in the model fitting. In tradeSeq we made the pragmatic decision to not do so. Please see #164 for a discussion on this.
We do recommend to work with the SCE output. I have to say I am not sure what you mean with the new prediction statement. Could you please clarify?

koenvandenberge commented

Closing due to inactivity. Feel free to reopen as needed.