Integrateing 10X Multiome

[AAA-3]

Hello,

I am trying to integrate 4 runs of my multiome experiment and I am unsure of how to proceed. Having looked at issue #5346 and discussion #438 here is what I did:

1. I created 4 SEPARATE multiome objects for ease of QC parametres

h9.data <- Read10X_h5('.../filtered_feature_bc_matrix.h5')
h9 <- CreateSeuratObject(counts = h9.data$`Gene Expression`, project = "h9")
h9[['percent.mt']] <- PercentageFeatureSet(h9, pattern = '^MT-')

h9[["ATAC"]] <- CreateChromatinAssay(
  counts = h9.data$Peaks,
  sep = c(":", "-"),
  fragments = h9.atac,
  annotation = annotation)

h9 <- subset(h9, subset = ...)

2. After QC, I ran Diet seurat on each of the 4 multiome to get either the RNA or ATAC seurat obj (this ensured for me that the cells remain the same in both obj)

h9.rna <- DietSeurat(h9_scrubQC, assays = "RNA") # repeat for atac

3. I integrated the RNA objects using the SCT method

h9rna.list <- list(h9, ...)
h9rna.list <- lapply(X = h9rna.list, FUN = SCTransform, vars.to.regress = "percent.mt")
h9.features <- SelectIntegrationFeatures(object.list = h9rna.list)
h9rna.list <- PrepSCTIntegration(object.list = h9rna.list, anchor.features = h9.features) 
h9rna.list <- lapply(X = h9rna.list, FUN = RunPCA, features = h9.features, npcs = 100)h9anchors25 <- FindIntegrationAnchors(object.list = h9rna.list, reference = c(1, 2), scale = FALSE, normalization.method = "SCT", reduction = "rpca", dims = 1:100, anchor.features = h9.features, k.anchor = 25)
h9RNA25 <- IntegrateData(anchorset = h9anchors25, normalization.method = "SCT", dims = 1:100) %>% RunPCA(npcs = 100) %>% RunUMAP(dims = 1:100)

4. I merged ATAC objects using the merged set of peaks (reduced common peaks) followed by integration

h9.peaks <- read.table(file = ".../atac_peaks.bed", col.names = c("chr", "start", "end"))
h9.gr <-  makeGRangesFromDataFrame(h9.peaks)
h9.combined.peaks <- reduce(x = c(h9.gr, ...), with.revmap = T)
h9.counts <- FeatureMatrix(fragments = frags.h9, features = h9.combined.peaks, cells = h9metrics.common$barcode) #h9metrics.common was created with a list of cells that had passed QC
h9_assay <- CreateChromatinAssay(h9.counts, fragments = frags.h9)
h9.atac <- CreateSeuratObject(h9_assay, assay = "ATAC", meta.data=h9metrics.common)

#processing the atac objects individually 
h9.atac <- FindTopFeatures(h9.atac) %>% RunTFIDF() %>% RunSVD(n = 100)

#merge: 
h9.combined <- merge(x = h9.atac, y = list(...))
h9.combined <- FindTopFeatures(h9.combined) %>% RunTFIDF() %>% RunSVD(n = 100) %>% RunUMAP(reduction = "lsi", dims = 2:100)

#integrate 
h9.integration.anchors10 <- FindIntegrationAnchors(object.list = list(h9.atac, ...), anchor.features = rownames(h9.combined), reduction = "rlsi", dims = 2:100, k.anchor = 10, reference = c(1,2))
h9.integrated10 <- IntegrateEmbeddings(anchorset = h9.integration.anchors10, reductions = h9.combined[["lsi"]],new.reduction.name = "integrated_lsi", dims.to.integrate = 1:100) %>% RunUMAP(reduction = "integrated_lsi", dims = 2:100)

Now I have 2 files - an integrated RNA and an integrated ATAC. but whenever I try to run RNA_obj[["ATAC"]]<-ATAC_obj it does not work

Error in `[[<-`: ! Cannot add new cells with [[<- Backtrace:  1. methods (local) `[[<-`(`*tmp*`, "ATAC", value = `<ChrmtnAs[,25183]>`)  2. SeuratObject (local) `[[<-`(`*tmp*`, "ATAC", value = `<ChrmtnAs[,25183]>`)

RNA_obj[["ATAC"]]<-ATAC_obj[["ATAC"]] it does not seem to work either - am I missing someting?

[AAA-3]

Nevermind! One of the objects I merged did not have matching cell id - sorry, should have checked before posting