Inclusion of extra-species motifs in RunChromVAR()

[sarah-chapin]

Hello,

I am running chromVAR via the Signac wrapper using the code below:

jaspar24 <- JASPAR2024()
sq24 <- RSQLite::dbConnect(RSQLite::SQLite(), db(jaspar24))
pfm24 <- TFBSTools::getMatrixSet(
  x = sq24,
  opts = list(collection = "CORE", tax_group = 'vertebrates', all_versions = FALSE)
)

atac <- AddMotifs(
  object = atac,
  genome = BSgenome.Hsapiens.UCSC.hg38,
  pfm = pfm24
)

main.chroms <- standardChromosomes(BSgenome.Hsapiens.UCSC.hg38)
keep.peaks <- as.logical(seqnames(granges(atac)) %in% main.chroms)
atac <- atac[keep.peaks, ]

# run chromvar
atac <- RunChromVAR(
  object = atac,
  genome = BSgenome.Hsapiens.UCSC.hg38
)

Although I am including the human genome when adding the motifs and running chromVAR, I still have some mouse motifs that are scored. Do you know why this would be? Furthermore, will the inclusion of the mouse motifs affect the scores for the human motifs?

In addition, when I assess differential motif activity via the code suggested in the Signac tutorial (shown below), I find that there are some celltypes with hundreds of differentially-active motifs. Is it typical to see this number of differentially-active motifs?

differential_activity <- FindAllMarkers(
  object = seu,
  only.pos = TRUE,
  mean.fxn = rowMeans,
  fc.name = "avg_diff",
  group.by = "celltype"
)

[timoast]

You are using the core vertebrate motif collection from JASPAR, so it is not limited to the human motifs. In chromVAR each motif is independent so it will not affect the other motifs.

I find that there are some celltypes with hundreds of differentially-active motifs. Is it typical to see this number of differentially-active motifs?

This depends on the cell types being compared and on the criteria for defining motifs as differential