Multimodal FindMultiModalNeighbors after Harmony on both GEX + ATAC data

[Gimenagomez2230]

Hi, I am trying to analyze Multiome 10X data following the vignette (https://stuartlab.org/signac/articles/pbmc_multiomic).

So far, I processed GEX (SCTransform each sample, merge, pca) and also processed ATAC data.

I harmonized both GEX and ATAC separate to account for batch and Clone, so apart from pca and lsi, I have harmonized reductions I want to use in my bimodal:

So instead of:
pbmc <- FindMultiModalNeighbors (
object = pbmc,
reduction.list = list("pca", "lsi"),
dims.list = list(1:50, 2:40),
modality.weight.name = "RNA.weight",
verbose = TRUE
)

I have:

seurat.joined.sct <- FindMultiModalNeighbors(seurat.joined.sct,
reduction.list = list("harmony_batch_clon", "harmony_atac_batch_clon"),
dims.list = list(1:ncol(Embeddings(seurat.joined.sct,"harmony_batch_clon")),
1:ncol(Embeddings(seurat.joined.sct,"harmony_atac_batch_clon"))),
modality.weight.name = "RNA.weight",
verbose = TRUE)

I get:
Warning message:
In FindMultiModalNeighbors(seurat.joined.sct, reduction.list = list("harmony_batch_clon", :
The number of provided modality.weight.name is not equal to the number of modalities. SCT.weight ATAC.weight are used to store the modality weights

1. I dont understand this, as in the vignette it also has two modalities but the modality.weight.name is one as well: "RNA.weight" ...

2. Given that I have SCTransform should I use "SCT.weight" instead of "RNA.weight" ? But again, in the vignette they also do SCTransform of RNA assay but here they use "RNA.weight" not "SCT.weight" .

3. Is this warning telling me that no matter what I put there, it will use anyway "SCT.weight", "ATAC.weight "? Its that fine?

4. If it needs two, should I use modality.weight.name = c("SCT.weight", "ATAC.weight ") or modality.weight.name = c("RNA.weight", "ATAC.weight ")?

Finally, I think my GEX data is much better that my ATAC data, so I was hoping that by saying just "RNA.weight"(or "SCT.weight") it would prioritize GEX data for UMAP and clustering...

Any advice is welcome!
Thank you so much for all your tools and help!

[timoast]

Since these questions relate to Seurat functions rather than Signac, I suggest you raise an issue there: https://github.com/satijalab/seurat