Corrected count for resolvi

Help scvi-tools
1

Hi,

I am wondering what would be the right way of generating a corrected count matrix for downstream analysis from resolvi.

The tutorial uses the median of px_rate as generated_expression, while the preprint computed corrected_counts by multiplying counts and px_rate_q25 / (1 + px_rate_q25 + mean_poisson).

From my understanding, the corrected_counts should be the right way to go, is it correct?

Thanks in advance!

2

Throughout the manuscript, I’m using p25 of ‚obs‘ using the corrected model, where I sample only from true expression and set diffusion and background to zero. I think this is a good starting point. In the tutorial, I’m outputting estimated counts using px_rate. If you are afraid of removing expression of some genes, the formula for corrected counts might be beneficial.