Results #611
-
|
Hello everyone! For example, pink cluster is the most differentiated one, and green is an intermediate state, so it does not make sense the vector's direction When I run the dataset with this guideline: https://cellrank.readthedocs.io/en/stable/beyond_rna_velocity.html I appreciate your help / opinions |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
|
@Marius1311 I just noticed in another discussion that you mentioned that you don't get notifications for new discussions. So I am just tagging you in case no one sees this! Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @pabloatria18! This looks interesting! I did in fact not get a notification for this - @michalk8, can you look into the settings please and change them so that we get notified when new questions are opened? Thanks! RNA velocity does in fact not seem to work very well on your data - the arrows point from most differentiated towards less differentiated. To figure out what the problem with the actual velocity vectors is, I would recommend opening a question in scVelo as that's not really related to CellRank per se. As to the second part of your analysis, please be aware of the fact that you're not using RNA velocity here - you're using CytoTRACE to infuse directionality into the KNN graph. I have never worked with CytoTRACE in a perturbation setting, so it's important that you read the original publication to get familiar with the biological assumptions made to check whether these hold for your system. Assuming for now that these assumptions are met, your trajectory analysis indicates that upon your perturbation ('experimental'), cells acquire a higher probability of transitioning towards cluster 5. You can quantify this fate bias by computing fate probabilities towards clusters 5 & 6, see e.g. the kernels and estimators tutorial. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you very much @Marius1311 ! Best ! |
Beta Was this translation helpful? Give feedback.
Hi @pabloatria18! This looks interesting! I did in fact not get a notification for this - @michalk8, can you look into the settings please and change them so that we get notified when new questions are opened? Thanks!
RNA velocity does in fact not seem to work very well on your data - the arrows point from most differentiated towards less differentiated. To figure out what the problem with the actual velocity vectors is, I would recommend opening a question in scVelo as that's not really related to CellRank per se.
As to the second part of your analysis, please be aware of the fact that you're not using RNA velocity here - you're using CytoTRACE to infuse directionality into the KNN graph…