countmatrix normalization

[karam-i]

Dear @wkopp

I'm trying to run scregseg on real bulk ATAC-seq data to have sample specific states following the tutorial for cluster collapsed tracks.
Does the model normalize the countmatix? or should it be normalized beforehand, and if so, which normalization method do you recommend?

Thank you in advance

Karam

[wkopp]

Hi Karam,
You don't need to normalize the count matrices. The multinomial emission probabilities would automatically account for the read depth per sample. That is, the emission probabilities should reflect read depth to some extend.
Best,
Wolfgang