parameters for scMACS::getDifferentialAccessibleTiles

[ziyuanhe]

Hey guys, I have been experimenting with the parameters in the getDifferentialAccessibleTiles functions. as explained to me by Samir, signalThreshold does appear to have a huge impact in the differential peaks detected.
in CD4 naive (7 preRA vs 6 healthy), if all other parameters are the same and same peak sets.

• signalThreshold=15 generates 12535 peaks at FDR<0.2
• signalThreshold=14 generates 13571 peaks at FDR<0.2
• signalThreshold=13 generates 7206 peaks at FDR<0.2
• signalThreshold=12 (default) generates 3400 peaks at FDR<0.2
• signalThreshold=10 generates 0 peaks at FDR<0.2

if i understand this line correctly idx <- which(medians_a > log2FC_filter | medians_b > log2FC_filter | diff0s >= minZeroDiff),
signalThreshold filters out the peaks whose median are lower than this value in both groups. What about filter by logfc difference as in seurat in scRNA differential or you are worried about the smaller peaks being noisy? even though higher threshold seems to give me more DAPs, my worries is that i will completely missing the smaller peaks which maybe on/off status.

any insights of how to set this parameter based on the data?

Here is the underline peak matrix distribution for CD4 naive
preRA_tea_seq_cd4tna_mocha_peaks_cell_type_histogram.pdf