WNN de novo UMAP from mapped scRNA and scATAC

[sabrinacamp2]

Hi,

I used the WNN vignette to create a WNN graph and UMAP for 10x multiome RNA + ATAC samples. For the FindMultiModalNeighbors function, the two reductions I am using are lsi for the ATAC assay and pca for the RNA assay.

I used the multimodal reference mapping vignette to map a single assay ATAC sample and a single assay RNA sample to the WNN UMAP and have the integrated LSI and PCA for these samples.

Is it possible to subset the WNN graph with the mapped samples and re-cluster? I saw where you could create a de novo UMAP for just one assay, .e.g. combining the reference and query PCs and creating a UMAP, but can you do this for both assays? I'm stuck at feeling like if I want to explore the cells beyond the inferred label from MapQuery, I need to consider each assay separately again instead of leveraging the WNN approach.

Best,
Sabrina