No Cluster ID's present when making UMAP for Joint RNA and ATAC analysis: Signac 10x multiomic tutorial

Hello,

I've been following the Joint RNA and ATAC analysis: 10x multiomic tutorial with my own snRNAseq + snATACseq data. I've completed all steps until the Joint UMAP visualization step. I skipped the annotating cell types step because I don't have annotations for this data yet, which is from the Primary Visual Cortex (V1) from Macaca fascicularis. I'm trying to create a UMAP for my 7 samples, but I'm not sure why it isn't generating a normal UMAP with cluster id's for each cluster. Below is my code for sample 1.

# build a joint neighbor graph using both assays
liftoff_1_MI5_V1_SO <- FindMultiModalNeighbors(
  object = liftoff_1_MI5_V1_SO,
  reduction.list = list("pca", "lsi"), 
  dims.list = list(1:50, 2:40),
  modality.weight.name = "RNA.weight",
  verbose = TRUE
)

# build a joint UMAP visualization
liftoff_1_MI5_V1_SO <- RunUMAP(
  object = liftoff_1_MI5_V1_SO,
  nn.name = "weighted.nn",
  assay = "RNA",
  verbose = TRUE
)

DimPlot(liftoff_1_MI5_V1_SO, label = TRUE, repel = TRUE, reduction = "umap") + NoLegend()

# not sure why everything is one color?

no cluster id umap

It seems to color all the cells by orig.ident which is not what I was expecting. I then tried the standard Seurat steps from their Guided Clustering Tutorial to generate the UMAP, but I was getting errors.

liftoff_1_MI5_V1_SO <- FindNeighbors(liftoff_1_MI5_V1_SO, dims = 1:30, return.neighbor = TRUE)
# Computing nearest neighbor graph
# Computing SNN

liftoff_1_MI5_V1_SO <- FindClusters(liftoff_1_MI5_V1_SO, resolution = c(0.4, 0.5, 0.6, 0.7, 0.8))
# Error in FindClusters.Seurat(liftoff_1_MI5_V1_SO, resolution = c(0.4,  : 
#   Provided graph.name not present in Seurat object

liftoff_1_MI5_V1_SO <- RunUMAP(liftoff_1_MI5_V1_SO, dims = 1:30)
DimPlot(liftoff_1_MI5_V1_SO, reduction = "umap", label = TRUE)

I'm not sure why I'm getting this error? Do I need to run the Seurat steps to generate the UMAP for each sample? Do I need to run the standard Seurat Find Neighbors, Find Clusters, RunUMAP, & DimPlot separately for the RNA assay and the ATAC assay before making a joint UMAP? How can I calculate the resolution for the joint (snRNAseq + snATACseq) UMAP? How can I make the UMAP so that each cluster has a cluster ID and each cluster has a different color like normal?

multiomics snatacseq snrnaseq

• 1,804 views • link

written 1.7 years ago by bgbs •

1 answer

Your object has no clusters set as Idents.

So yes, you need to call clusters first.

liftoff_1_MI5_V1_SO <- FindNeighbors(liftoff_1_MI5_V1_SO, dims = 1:30, return.neighbor = TRUE)

Remove return.neighbor=TRUE, not sure why you're using that.

liftoff_1_MI5_V1_SO <- FindClusters(liftoff_1_MI5_V1_SO, resolution = c(0.4, 0.5, 0.6, 0.7, 0.8))

Only 1 resolution should be provided.

• 0 views • link

written 1.7 years ago by jared.andrews07

Thanks for your response. I was able to fix this by first making the UMAP for the snRNAseq assay. I re-ran FindNeighbors and then re-ran FindClusters after specifying the neighbor method to use to find clusters with the graph.name parameter.

liftoff_1_MI5_V1_SO <- FindNeighbors(liftoff_1_MI5_V1_SO, dims = 1:30)
# > names(liftoff_1_MI5_V1_SO@graphs)
# [1] "wknn"    "wsnn"    "SCT_nn"  "SCT_snn"
# try running this find clusters command (WORKED!)
liftoff_1_MI5_V1_SO <- FindClusters(liftoff_1_MI5_V1_SO, graph.name = "SCT_snn", resolution = 0.4)
# "SCT_snn" is used as it corresponds to the shared nearest neighbor (SNN) graph generated during normalization with SCTransform
liftoff_1_MI5_V1_SO <- RunUMAP(liftoff_1_MI5_V1_SO, dims = 1:30)
DimPlot(liftoff_1_MI5_V1_SO, reduction = "umap", group.by = "SCT_snn_res.0.4", label = TRUE)

Next, I made a separate UMAP for the snATACseq assay.

# I already ran SCTransform (snRNAseq) and RunTFIDF, FindTopFeatures, & RunSVD (snATACseq) for each sample
# ATAC analysis
# We exclude the first dimension as this is typically correlated with sequencing depth
liftoff_1_MI5_V1_SO <- RunUMAP(liftoff_1_MI5_V1_SO, reduction = 'lsi', dims = 2:50, reduction.name = "umap.atac", reduction.key = "atacUMAP_")
# reduction.key is what the x and y axis are labeled by (PC1 ON X AND PC2 ON Y)
DimPlot(liftoff_1_MI5_V1_SO, reduction = "umap.atac", label = TRUE, label.size = 2.5, repel = TRUE) + ggtitle("1_MI5_snATAC")

Lastly, I made a combined multiomics UMAP (snRNAseq + snATACseq) using the following tutorial.

# resolution is 0.4
liftoff_1_MI5_V1_SO <- FindMultiModalNeighbors(liftoff_1_MI5_V1_SO, reduction.list = list("pca", "lsi"), dims.list = list(1:30, 2:30))
liftoff_1_MI5_V1_SO <- RunUMAP(liftoff_1_MI5_V1_SO, nn.name = "weighted.nn", reduction.name = "wnn.umap", reduction.key = "wnnUMAP_")
liftoff_1_MI5_V1_SO <- FindClusters(liftoff_1_MI5_V1_SO, graph.name = "wsnn", algorithm = 3, verbose = FALSE)
DimPlot(liftoff_1_MI5_V1_SO, reduction = "wnn.umap", label = TRUE, label.size = 2.5, repel = TRUE) + ggtitle("snRNA + snATAC")

• 0 views • link

written 1.7 years ago by bgbs •