This is a test version of Biostars. For the public version, visit https://www.biostars.org.
RNA velocity plot demystified

Here is an example of RNA velocity from Manno .et .al 2018.

enter image description here

Colour of cells denote cell type. Layout denotes how cells cluster by some measure of similarity. The direction of the arrows denote the future state of the current cells. Or I guess one can say the differentiation trajectory.

What do length of the arrows denote? Can we say something about the propensity of that cell to differentiate in that direction? Can we say if there are more or less spliced or unspliced transcripts?

My understanding is that an arrow is based on the number of spliced and unspliced transcripts for each gene, summarised over all genes and further summarised over all cells in a small neighbourhood.

Other comments on understanding RNA velocity plots are welcome. I am trying to understand this better.

rna-velocity velocyto single-cell

2 answers

This paragraph in Svensson and Pachter, Molecular Cell 2018 (https://www.sciencedirect.com/science/article/pii/S1097276518307974) might help understand the concept of these arrows:

"Long arrows, corresponding to large changes in gene expression, will in the low-dimensional representation imply that cells are undergoing rapid differentiation, while short arrows can be intuitively associated with terminally differentiated cells maintaining homeostasis. While some care needs to be taken when interpreting the lengths of arrows in low-dimensional representations, RNA velocity representations can be biologically informative. In principal component analysis, the vector lengths are meaningful, scaling linearly with RNA velocity estimates in high dimensions. We note that this will not be the case in t-SNE representations (Van der Maaten and Hinton, 2008), though directions remain meaningful and can be interpreted as “RNA arguments” in vector nomenclature."

There is a notebook available with the code to generate that plot if you want to understand the specific steps.

The method finds RNA "velocity" based on spliced and unspliced transcripts. The ratio is predictive of cellular state progression. The arrows show that.

Thanks for that. That might be helpful even though I don't understand python code. And the R notebooks unfortunately don't seem to go into so much detail. Anyhow, my question still stands as to what the LENGTH of the arrows mean.

Log in to answer this question.