Is Pseudotime analysis most applicable to time-series experiments?

Question

[Payal]

Hi,

This is a theoretical question. Is pseudotime analysis generally applied to time-series type experiment? Or can also be applied for a single time point.

Thanks, Payal

[piyushjo]

A single time point sample can also have cells from different developmental stages. For example, if you have sample of brain and you dissected region including SVZ to cortical layer, you will have the developmental stages a cell will pass through its differentiation gene. In that case a trajectory analysis will be useful.

A contrasting case would be, if you have HeLa cells growing in petri dish without any extra experimental manipulation, you are not expecting it to have any temporal stages (except cell cycle), so there doing trajectory analysis wouldn't be that beneficial.

So a little idea about the biology of sample to see if trajectory analysis is really needed or not. You will get trajectory not matter what, because of the heterogeneity.

[Payal]

Thanks @piyushjo. I am working on a brain sample from SVZ region only at a single timepoint. I am using Monocle3 and am getting disjointed trajectories because there are so many different types of populations in the sample.

Answer 1

[Jean-Karim Heriche]

It depends on what you mean by apply and in which context.
If you already have generated a trajectory or pseudotime reference, you can try and project individual points onto it.
If your single time point data consists of a mixture of "stages" because these happen asynchronously in your sample (e.g. mixture of cell cycle phases in a cell population), you could try to recover the stages and impose an ordering to build a trajectory.
Note that this is closely related to what is called seriation or ordination in other fields.