Preprocessing of mouse retinal cells

[WeilerP]

@sbooeshaghi, I went through your notebook on the mouse retinal cell analysis. During the normalization of the data using scv.pp.normalize_per_cell(adata), scVelo is throwing the warning WARNING: Did not normalize X as it looks processed already. To enforce normalization, setenforce=True. (same for layers unspliced and spliced). I had a look at the data and this is caused by the counts not being integers. Are the reads that kallisto bustools provides pseudo counts?

[sbooeshaghi]

kallisto | bustools will return non-integer counts >=0 if ambiguous 3' reads are disambiguated with the EM algorithm. Practically speaking this happens if bustools count --em is chosen or if bustools count without --genecounts is selected and then kallisto quant-tcc is ran to estimate isoform abundances from TCC counts