Load packages

library(LinQView)

Load data and create object

# Load in the RNA UMI matrix for PBMC 1K dataset
cbmc.1k.data <- readDataFrom10X(dir = "../1K/filtered_feature_bc_matrix/")
cbmc.1k <- createObject(data = cbmc.1k.data, project = "CBMC1K")
# Load in the RNA UMI matrix for PBMC 10K dataset
cbmc.10k.data <- readDataFrom10X(dir = "../10K/filtered_feature_bc_matrix/")
cbmc.10k <- createObject(data = cbmc.10k.data, project = "CBMC10K")

pre-process for two datasets

cbmc.1k <- dataNormalization(object = cbmc.1k)
cbmc.1k <- FindVariableFeatures(cbmc.1k, selection.method = "vst", nfeatures = 2000)

cbmc.10k <- dataNormalization(object = cbmc.10k)
cbmc.10k <- FindVariableFeatures(cbmc.10k, selection.method = "vst", nfeatures = 2000)

Integrate two dataset using Seurat V3 (integrate RNA expression)

cbmc.anchors <- FindIntegrationAnchors(object.list = list(cbmc.1k, cbmc.10k), 
    dims = 1:20)
cbmc.integrated <- IntegrateData(anchorset = cbmc.anchors, dims = 1:20)

Joint PCA

DefaultAssay(object = cbmc.integrated) <- "integrated"
# PCA
cbmc.integrated <- dataScaling(object = cbmc.integrated)
cbmc.integrated <- RunPCA(cbmc.integrated, verbose = FALSE, features = VariableFeatures(object = cbmc.integrated))

Joint RNA and ADT distances, clusters, UMAPs

# do distance, clustering and umaps in the same step cbmc.integrated <-
# jointClusteing(object = cbmc.integrated, run_seprate = TRUE)

# joint distance
cbmc.integrated <- jointDistance(cbmc.integrated)
# umaps
cbmc.integrated <- umapFromDistane(cbmc.integrated)
# clustering
cbmc.integrated <- clusteringFromDistance(cbmc.integrated)

directly pool two datasets together as comparision

cbmc.merge <- merge(x = cbmc.1k, y = cbmc.10k)
cbmc.merge <- FindVariableFeatures(object = cbmc.merge)
cbmc.merge <- ScaleData(object = cbmc.merge)
cbmc.merge <- RunPCA(object = cbmc.merge)
# joint distance
cbmc.merge <- jointDistance(cbmc.merge)
# umaps
cbmc.merge <- umapFromDistane(cbmc.merge)
# clustering
cbmc.merge <- clusteringFromDistance(cbmc.merge)

cells from two datasets after Batch effects correction on RNA, ADT and joint map

p01 <- DimPlot(cbmc.merge, reduction = "umap_rna", group.by = "orig.ident") + 
    ggtitle("RNA UMAP without Batch Effects Correction")
p02 <- DimPlot(cbmc.merge, reduction = "umap_adt", group.by = "orig.ident") + 
    ggtitle("ADT UMAP without Batch Effects Correction")
p03 <- DimPlot(cbmc.merge, reduction = "umap_joint", group.by = "orig.ident") + 
    ggtitle("Joint UMAP without Batch Effects Correction")
p1 <- DimPlot(cbmc.integrated, reduction = "umap_rna", group.by = "orig.ident") + 
    ggtitle("RNA UMAP with Batch Effects Correction")
p2 <- DimPlot(cbmc.integrated, reduction = "umap_adt", group.by = "orig.ident") + 
    ggtitle("ADT UMAP with Batch Effects Correction")
p3 <- DimPlot(cbmc.integrated, reduction = "umap_joint", group.by = "orig.ident") + 
    ggtitle("Joint UMAP with Batch Effects Correction")
plot_grid(p01, p02, p03, p1, p2, p3, ncol = 3)

visualization ADT vs RNA vs Joint

plots <- generateGridDimPlot(cbmc.integrated, legend = FALSE, reduction.prefix = "umap_", 
    darkTheme = FALSE)
listPlot(object = plots)

Heat map that show both ADT and RNA

# Heatmap for joint clusters
heatMapPlot(object = cbmc.integrated, group.by = "jointClusterID")