Differential accessibility, conditions and replicates

[gouthamatla]

Hi,

I am trying to get my head around the differential accessibility analysis across three conditions, using replicate samples.

I have two human samples that are split into three parts and exposed to different "conditions" followed by scATAC experiment.

I merged all the data into a single Signac object. I could see cells cluster based on cell-types but there are sample specific effects (I.e cells cluster based on cell-type but separated by sample, which is expected). Now I am interested to understand the effect of "condition" on different cell types (let's say cell type A). Naively I did a differential analysis on cell-type A between pairs of conditions using all the cells together. Here, obvious problem is that I am not using the replicate information to gain statistical power I.e I am not assessing if the response of cell type A is consistent between two samples.

I have read about using latent.vars in FinaMarkers function. Should I provide the sample IDs to this function to make use of replicates in statistical analysis to ensure only the consistent results in both replicates are reported ?

Other option is to analyze both samples separately and ask which regions are differentially accessible (same direction) in both the samples.

I Would like to know which would be a better way to make use of replicate samples.

Hope my question is clear.

Thanks in advance,