Analyze the umap trajectory

[dhoneyi]

Hi, Thanks for your great works and sharing your code. When I use your analyze the umap trajectory code to construct the trajectory in my own data, I have some questions. Hope you can help me figure out. I feel a little confused that if I have no idea the development background in my data, how can I choose the clusters to build the trajectory? In the paper, you extract the B cell associated cluster to reverse reconstruction of B cell differentiation trajectory. I wanna know can I use the unsupervised methods to learn the trajectory. Some specific questions for your code, how you choose the latest cluster? And which matrix contain the delta ATAC number between each cluster? Thanks a lot.