Hi!

I'm using chromVAR on scATAC-seq data for 6 different samples, corresponding to 3 different genotypes (WT and 2 knock-outs). My count matrix consists of 16,000 cells (for the 6 samples) and 137,000 peaks (peaks shared by all cells from my different samples). As the job was taking too much memory, I ran chomVAR 3 times for the 3 different genotypes and then concatenated the results. For each job, I used the addGCBias function to create a background for the peaks. I checked the 3 distributions of the TF deviation scores for each motif and they seem to overlap. Does it make sense to compare the scores across genotypes or is there a cofounder that I may not take into account here?

Thank you so much for your help.