Specifying start / end cell (Specify root_cells and end_point)

Hi,

Many thanks for the scVeloo toolkit and the github tutorials. Pretaining to issue #157 #112, I also encountered that the direction goes opposite to what is expected (Day 7 > Day 5 > Day 3) rather than (Day 3 > Day 5 > Day 7). I tried some of the suggestion mentioned in the others thread mainy the different mode steady_state, stochastic, dynamical, but it gives very simillar results.

In one of the comments @VolkerBergen you mentioned

The roots may have already been computed e.g. via tl.terminal_states or given by the user via adata.uns[root_key], otherwise it is computed here internally.

So at the moment, both pseudotime (Figure below ‘velocity pseudotime’) and direction is going in the opposite direction. Therefore I was wondering if its possible to set the root_cells? And would this hopefully invert the peudotime and direction…

scv.pl.scatter(mdata, color='velocity_pseudotime'
scv.tl.terminal_states(mdata)
scv.pl.scatter(mdata, color=['root_cells', 'end_points''])

In #114 you did mention

We will work out a more robust way of confidently assigning root cells soon. For now, you can either store a root cell inadata.uns['iroot']and use root_key=‘iroot’, or you can constrain the adata.obs['root_cells'] to the bottom cells.

Does the iroot refer to one cell or a cluster/group of cells?

Best Zaki

[hyjforesight]

Hello scVelo, Could you please tell us how to set the root clusters for running scVelo? My issue is that the initial cluster calculated by scVelo internally is not consistent with the classical Mki67+ and Lgr5+ clusters, which are already known as the stem cell clusters. Thanks! Best, YJ

[ccruizm]

Good day!

I was wondering if there is a solution for these datasets where the velocity is inverted. Thanks !