Differences in cell proportions between conditions between scRNA-seq and scATAC-seq

[DaianeH]

Hi,

We have a cohort of AML patients, some with mutations on TP53 and some without (WT), and we'd like to find transcriptional differences between the two conditions. scRNA-seq and scATAC-seq were generated on the same patients. I followed archR tutorial to integrate the two datasets and transfer labels from scRNA-seq to scATAC-seq.

I would like to know if there's a difference in proportion of cells between the two groups.

In scRNA-seq, patients with TP53 MUT seem to have more Late Erythroid and LSC cells. However, that is not replicated on the scATAC-seq dataset, where results are completely different: TP53 MUT patients seem to have more Memory B, NK and Plasma. There's so few LSC cells in scATAC-seq that I had to exclude this cell-type from the table to be able to generate the plot without errors.

I was expecting to be able to replicate the proportion of cells between the two experiments, since they were performed in the same patients and labels were transferred from one dataset to another. Does anyone have an explanation as to why I'm not seeing this?

Thank you so much in advance,