Entering edit mode
15 months ago
Chironex
▴
50
hi, I ran the analysis of my dataset with monocle3 twice: the first on WT cells, the second on mutant cells. I wonder if there was a method to compare the differences between the two trajectories (wild type and mutant) and make sure that they are not the result of artifacts. obviously I produced two umaps to compare.
Suggestions about other tools are appreciated. I wasn't able to find something that fit my needs
Hi, I haven't used Monocle, but similar dataset with other tools. Why not process the WT and mut. together, for UMAP? Once done, you can then see which (WT/ Mut) is more in which cluster, or which cluster is more in WT/ Mut. I have used Phate as a trajectory method. Once you have the trajectory, you can then overlay the UMAP cluster info. and check if the clusters more in Mut. are towards the end of the (pseudotime) trajectory [i.e., one possible biological scenario]
Hi, thank you for the response, I never used it for analysis. Do you have any suggestions to do this? I have a Seurat object with umap, pca, etc. I should let phate produce its own umap? thank you
Phate (or Monocle) doesn't do UMAP. If you have done UMAP already (using seurat) then you could use the cell barcode to UMAP cluster label coding as colour assignment on the Phate trajectory. Notice how the colour argument is being specified in the Phate plots in this tutorial here. Instead of using any gene's expression value as colour code (like in the tutorial; e.g. see 4th plot from last), you could use any other value that has one-to-one mapping for the cell barcodes (like seurat cluster label).
Though its not necessary to overlay seurat-umap cluster labels onto a pseudotime trajectory output like of Phate, but doing so allows you to check if the biology makes sense. Cells which appear towards the end point of the pseudotime trajectory plot could be indicative of a differentiated cell state. If those cells on the UMAP plot are mostly belonging to clusters that are expressing differentiation marker genes, then it could be said that the two methods are probably picking up the same trend.