Pancreas basic tutorial - Gene expression trends: Var annotations not found

[le-ander]

I'm trying out some example from the notebook and got stuck in the gene dynamics heatmap section right at the bottom of the turoial. In my adata object, I cannot find the key adata.var['to TERMINAL_STATE'] (TERMINAL_STATE being Alpha in the tutorial. In my adata.var I only have the columns to TERMINAL_STATE corr and to TERMINAL_STATE qval

Did I maybe miss a step in the pipeline that would have created the missing column?

This is the section I am referring to:

[michalk8]

Hi @le-ander , you're right, the tutorial notebooks are slightly outdated (my CI should've let me know, but no more Travis credits). I will regenerate it now (btw I recommend the examples, since they are always up-to-date). As for the keys, in https://github.com/theislab/cellrank/pull/437, to TERMINAL_STATE was rename to TERMINAL_STATE corr to better distinguish the columns (like the pval or qval or ci low and ci high).

[le-ander]

Thanks for the clarification! I guess now that you have the qvalues readily available as well you could also use the top 100 genes by qvalue, which would give you a mix of positively and negatively correlated genes.

I find the negatively correlated ones also interesting as the turning off of certain genes can also be considered as a lineage definition. Any thoughts on this? also @Marius1311

[Marius1311]

I think for the tutorial it's fine as it stands because we're only interested in showing the functionality, but you're right, for general applications, it can be quite interesting to look at down-regulation as well.

We don't really do any sorting at the moment, we just add the key to .var, so it's really up to the user which genes they are interested in .

[michalk8]

I'm assuming this is has been solved (the tutorials have been regenerated), feel free to reopen.