How to determine optimum number of trajectories/diffusion times?

[sc30301]

Hi,

I am wondering how to determine/evaluate the ideal N and t.list parameters in the ExtractGeneTrajectory step? For example, in the tutorial, N=3 and t.list = c(4,7,7) were chosen, but how were these values selected?

[fra-pcmgf]

Hi @sc30301,

Adjusting the parameters can be a bit tricky and require some trial and error. The easiest way is to run the tool in a Jupyter notebook and use the ExtractGeneTrajectoryWidget to tweak parameters. There is an example notebook at https://github.com/KlugerLab/GeneTrajectory-python/blob/main/notebooks/tutorial_human_myeloid.ipynb

Typically the steps might include

• Consider adjusting k if running on a small dataset (e.g. the tutorial above)
• Look at the diffusion maps to determine the overall number of trajectories
• tweak values of t_list for each trajectory to make the trajectory longer / shorter.