Does a low unspliced proportion have an impact on scvelo analysis?

[jiangpuxuan]

Hi, I have many germ cell scRNA datasets, and all of them have a very low unspliced proportion ( about 5% to 10% ).
The scvelo+cellrank pipeline gave me a wierd result of terminal/initial state that is contrary to Prior Knowledge.
So, I wonder if a low unspliced proportion have an impact on scvelo analysis.
Thank you very much.

[WeilerP]

@jiangpuxuan, yes, it can (and most likely does) have an impact. Please have a look at related issues and discussions. My best guess guess is that something similar to #216 or #462 is happening.