Analysis of 10x PBMC 4k dataset

Keita Iida

2022-07-14

1 Computational environment

MacBook Pro (Big Sur, 16-inch, 2019), Processor (2.4 GHz 8-Core Intel Core i9), Memory (64 GB 2667 MHz DDR4).


2 Install libraries

Attach necessary libraries:

library(ASURAT)
library(SingleCellExperiment)
library(SummarizedExperiment)


3 Introduction

In this vignette, we analyze single-cell RNA sequencing (scRNA-seq) data obtained from peripheral blood mononuclear cells (PBMCs) of healthy donors.


4 Prepare scRNA-seq data

4.1 PBMC 4k

The data can be loaded by the following code:

pbmc <- readRDS(url("https://figshare.com/ndownloader/files/34112459"))

The data are stored in DOI:10.6084/m9.figshare.19200254 and the generating process is described below.


The data were obtained from 10x Genomics repository (PBMC 4k). Create SingleCellExperiment objects by inputting raw read count tables.

path_dir <- "rawdata/2020_001_10xgenomics/pbmc_4k/"
path_dir <- paste0(path_dir, "filtered_gene_bc_matrices/GRCh38/")
mat <- Seurat::Read10X(data.dir = path_dir, gene.column = 2,
                       unique.features = TRUE, strip.suffix = FALSE)
pbmc <- SingleCellExperiment(assays = list(counts = as.matrix(mat)),
                             rowData = data.frame(gene = rownames(mat)),
                             colData = data.frame(sample = colnames(mat)))
dim(pbmc)
[1] 33694  4340
# Save data.
saveRDS(pbmc, file = "backup/04_001_pbmc4k_rawdata.rds")

# Load data.
pbmc <- readRDS("backup/04_001_pbmc4k_rawdata.rds")


5 Preprocessing

5.1 Control data quality

Remove variables (genes) and samples (cells) with low quality, by processing the following three steps:

  1. remove variables based on expression profiles across samples,
  2. remove samples based on the numbers of reads and nonzero expressed variables,
  3. remove variables based on the mean read counts across samples.

First of all, add metadata for both variables and samples using ASURAT function add_metadata().

pbmc <- add_metadata(sce = pbmc, mitochondria_symbol = "^MT-")


5.1.1 Remove variables based on expression profiles

ASURAT function remove_variables() removes variable (gene) data such that the numbers of non-zero expressing samples (cells) are less than min_nsamples.

pbmc <- remove_variables(sce = pbmc, min_nsamples = 10)


5.1.2 Remove samples based on expression profiles

Qualities of sample (cell) data are confirmed based on proper visualization of colData(sce).

title <- "PBMC 4k"
df <- data.frame(x = colData(pbmc)$nReads, y = colData(pbmc)$nGenes)
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df$x, y = df$y), size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "Number of reads", y = "Number of genes") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 20))
filename <- "figures/figure_04_0010.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5, height = 5)

df <- data.frame(x = colData(pbmc)$nReads, y = colData(pbmc)$percMT)
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df$x, y = df$y), size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "Number of reads", y = "Perc of MT reads") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 20))
filename <- "figures/figure_04_0011.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5, height = 5)

ASURAT function remove_samples() removes sample (cell) data by setting cutoff values for the metadata.

pbmc <- remove_samples(sce = pbmc, min_nReads = 2000, max_nReads = 20000,
                       min_nGenes = 100, max_nGenes = 1e+10,
                       min_percMT = 0, max_percMT = 10)


5.1.3 Remove variables based on the mean read counts

Qualities of variable (gene) data are confirmed based on proper visualization of rowData(sce).

title <- "PBMC 4k"
aveexp <- apply(as.matrix(assay(pbmc, "counts")), 1, mean)
df <- data.frame(x = seq_len(nrow(rowData(pbmc))),
                 y = sort(aveexp, decreasing = TRUE))
p <- ggplot2::ggplot() + ggplot2::scale_y_log10() +
  ggplot2::geom_point(ggplot2::aes(x = df$x, y = df$y), size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "Rank of genes", y = "Mean read counts") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 20))
filename <- "figures/figure_04_0015.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5, height = 5)

ASURAT function remove_variables_second() removes variable (gene) data such that the mean read counts across samples are less than min_meannReads.

pbmc <- remove_variables_second(sce = pbmc, min_meannReads = 0.05)
dim(pbmc)
[1] 6568 4060
# Save data.
saveRDS(pbmc, file = "backup/04_002_pbmc4k_dataqc.rds")

# Load data.
pbmc <- readRDS("backup/04_002_pbmc4k_dataqc.rds")


5.2 Normalize data

Perform bayNorm() (Tang et al., Bioinformatics, 2020) for attenuating technical biases with respect to zero inflation and variation of capture efficiencies between samples (cells).

bayout <- bayNorm::bayNorm(Data = assay(pbmc, "counts"), mode_version = TRUE)
assay(pbmc, "normalized") <- bayout$Bay_out

Perform log-normalization with a pseudo count.

assay(pbmc, "logcounts") <- log(assay(pbmc, "normalized") + 1)

Center row data.

mat <- assay(pbmc, "logcounts")
assay(pbmc, "centered") <- sweep(mat, 1, apply(mat, 1, mean), FUN = "-")

Set gene expression data into altExp(sce).

sname <- "logcounts"
altExp(pbmc, sname) <- SummarizedExperiment(list(counts = assay(pbmc, sname)))

Add ENTREZ Gene IDs to rowData(sce).

dictionary <- AnnotationDbi::select(org.Hs.eg.db::org.Hs.eg.db,
                                    key = rownames(pbmc),
                                    columns = "ENTREZID", keytype = "SYMBOL")
dictionary <- dictionary[!duplicated(dictionary$SYMBOL), ]
rowData(pbmc)$geneID <- dictionary$ENTREZID
# Save data.
saveRDS(pbmc, file = "backup/04_003_pbmc4k_normalized.rds")

# Load data.
pbmc <- readRDS("backup/04_003_pbmc4k_normalized.rds")


6 Multifaceted sign analysis

Infer cell or disease types, biological functions, and signaling pathway activity at the single-cell level by inputting related databases.

ASURAT transforms centered read count tables to functional feature matrices, termed sign-by-sample matrices (SSMs). Using SSMs, perform unsupervised clustering of samples (cells).


6.1 Compute correlation matrices

Prepare correlation matrices of gene expressions.

mat <- t(as.matrix(assay(pbmc, "centered")))
cormat <- cor(mat, method = "spearman")
# Save data.
saveRDS(cormat, file = "backup/04_003_pbmc4k_cormat.rds")

# Load data.
cormat <- readRDS("backup/04_003_pbmc4k_cormat.rds")


6.2 Load databases

Load databases.

urlpath <- "https://github.com/keita-iida/ASURATDB/blob/main/genes2bioterm/"
load(url(paste0(urlpath, "20201213_human_CO.rda?raw=TRUE")))         # CO
load(url(paste0(urlpath, "20220308_human_MSigDB.rda?raw=TRUE")))     # MSigDB
load(url(paste0(urlpath, "20220308_human_CellMarker.rda?raw=TRUE"))) # CellMarker
load(url(paste0(urlpath, "20201213_human_GO_red.rda?raw=TRUE")))     # GO
load(url(paste0(urlpath, "20201213_human_KEGG.rda?raw=TRUE")))       # KEGG

The reformatted knowledge-based data were available from the following repositories:

Create a custom-built cell type-related databases by combining different databases for analyzing human single-cell transcriptome data.

d <- list(human_CO[["cell"]], human_MSigDB[["cell"]], human_CellMarker[["cell"]])
human_CB <- list(cell = do.call("rbind", d))

Add formatted databases to metadata(sce)$sign.

pbmcs <- list(CB = pbmc, GO = pbmc, KG = pbmc)
metadata(pbmcs$CB) <- list(sign = human_CB[["cell"]])
metadata(pbmcs$GO) <- list(sign = human_GO[["BP"]])
metadata(pbmcs$KG) <- list(sign = human_KEGG[["pathway"]])


6.3 Create signs

ASURAT function remove_signs() redefines functional gene sets for the input database by removing genes, which are not included in rownames(sce), and further removes biological terms including too few or too many genes.

pbmcs$CB <- remove_signs(sce = pbmcs$CB, min_ngenes = 2, max_ngenes = 1000)
pbmcs$GO <- remove_signs(sce = pbmcs$GO, min_ngenes = 2, max_ngenes = 1000)
pbmcs$KG <- remove_signs(sce = pbmcs$KG, min_ngenes = 2, max_ngenes = 1000)

ASURAT function cluster_genes() clusters functional gene sets using a correlation graph-based decomposition method, which produces strongly, variably, and weakly correlated gene sets (SCG, VCG, and WCG, respectively).

set.seed(1)
pbmcs$CB <- cluster_genesets(sce = pbmcs$CB, cormat = cormat,
                             th_posi = 0.20, th_nega = -0.40)
set.seed(1)
pbmcs$GO <- cluster_genesets(sce = pbmcs$GO, cormat = cormat,
                             th_posi = 0.20, th_nega = -0.30)
set.seed(1)
pbmcs$KG <- cluster_genesets(sce = pbmcs$KG, cormat = cormat,
                             th_posi = 0.20, th_nega = -0.20)

ASURAT function create_signs() creates signs by the following criteria:

  1. the number of genes in SCG>= min_cnt_strg (the default value is 2) and
  2. the number of genes in VCG>= min_cnt_vari (the default value is 2),

which are independently applied to SCGs and VCGs, respectively.

pbmcs$CB <- create_signs(sce = pbmcs$CB, min_cnt_strg = 4, min_cnt_vari = 4)
pbmcs$GO <- create_signs(sce = pbmcs$GO, min_cnt_strg = 3, min_cnt_vari = 3)
pbmcs$KG <- create_signs(sce = pbmcs$KG, min_cnt_strg = 3, min_cnt_vari = 3)


6.4 Select signs

If signs have semantic similarity information, one can use ASURAT function remove_signs_redundant() for removing redundant sings using the semantic similarity matrices.

simmat <- human_GO$similarity_matrix$BP
pbmcs$GO <- remove_signs_redundant(sce = pbmcs$GO, similarity_matrix = simmat,
                                   threshold = 0.85, keep_rareID = TRUE)

ASURAT function remove_signs_manually() removes signs by specifying IDs (e.g., GOID:XXX) or descriptions (e.g., metabolic) using grepl().

keywords <- "Covid|COVID"
pbmcs$KG <- remove_signs_manually(sce = pbmcs$KG, keywords = keywords)


6.5 Create sign-by-sample matrices

ASURAT function create_sce_signmatrix() creates a new SingleCellExperiment object new_sce, consisting of the following information:

pbmcs$CB <- makeSignMatrix(sce = pbmcs$CB, weight_strg = 0.5, weight_vari = 0.5)
pbmcs$GO <- makeSignMatrix(sce = pbmcs$GO, weight_strg = 0.5, weight_vari = 0.5)
pbmcs$KG <- makeSignMatrix(sce = pbmcs$KG, weight_strg = 0.5, weight_vari = 0.5)


6.6 Reduce dimensions of sign-by-sample matrices

Perform t-distributed stochastic neighbor embedding.

for(i in seq_along(pbmcs)){
  set.seed(1)
  mat <- t(as.matrix(assay(pbmcs[[i]], "counts")))
  res <- Rtsne::Rtsne(mat, dim = 2, pca = TRUE, initial_dims = 50)
  reducedDim(pbmcs[[i]], "TSNE") <- res[["Y"]]
}

Perform Uniform Manifold Approximation and Projection.

for(i in seq_along(pbmcs)){
  set.seed(1)
  mat <- t(as.matrix(assay(pbmcs[[i]], "counts")))
  res <- umap::umap(mat, n_components = 2)
  reducedDim(pbmcs[[i]], "UMAP") <- res[["layout"]]
}

Show the results of dimensional reduction in low-dimensional spaces.

titles <- c("PBMC 4k (cell type)", "PBMC 4k (function)", "PBMC 4k (pathway)")
for(i in seq_along(titles)){
  df <- as.data.frame(reducedDim(pbmcs[[i]], "UMAP"))
  p <- ggplot2::ggplot() +
    ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2]),
                        color = "black", size = 1, alpha = 1) +
    ggplot2::labs(title = titles[i], x = "UMAP_1", y = "UMAP_2") +
    ggplot2::theme_classic(base_size = 20) +
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18))
  filename <- sprintf("figures/figure_04_%04d.png", 19 + i)
  ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 4.1, height = 4.3)
}

# Save data.
saveRDS(pbmcs, file = "backup/04_004_pbmcs4k_ssm.rds")

# Load data.
pbmcs <- readRDS("backup/04_004_pbmcs4k_ssm.rds")


6.7 Cluster cells

6.7.1 Use Seurat functions

To date (December, 2021), one of the most useful clustering methods in scRNA-seq data analysis is a combination of a community detection algorithm and graph-based unsupervised clustering, developed in Seurat package.

Here, our strategy is as follows:

  1. convert SingleCellExperiment objects into Seurat objects (note that rowData() and colData() must have data),
  2. perform ScaleData(), RunPCA(), FindNeighbors(), and FindClusters(),
  3. convert Seurat objects into temporal SingleCellExperiment objects temp,
  4. add colData(temp)$seurat_clusters into colData(sce)$seurat_clusters.
resolutions <- c(0.15, 0.15, 0.15)
dims <- list(seq_len(40), seq_len(40), seq_len(10))
for(i in seq_along(pbmcs)){
  surt <- Seurat::as.Seurat(pbmcs[[i]], counts = "counts", data = "counts")
  mat <- as.matrix(assay(pbmcs[[i]], "counts"))
  surt[["SSM"]] <- Seurat::CreateAssayObject(counts = mat)
  Seurat::DefaultAssay(surt) <- "SSM"
  surt <- Seurat::ScaleData(surt, features = rownames(surt))
  surt <- Seurat::RunPCA(surt, features = rownames(surt))
  surt <- Seurat::FindNeighbors(surt, reduction = "pca", dims = dims[[i]])
  surt <- Seurat::FindClusters(surt, resolution = resolutions[i])
  temp <- Seurat::as.SingleCellExperiment(surt)
  colData(pbmcs[[i]])$seurat_clusters <- colData(temp)$seurat_clusters
}

Show the clustering results in low-dimensional spaces.

titles <- c("PBMC 4k (cell type)", "PBMC 4k (function)", "PBMC 4k (pathway)")
for(i in seq_along(titles)){
  labels <- colData(pbmcs[[i]])$seurat_clusters
  df <- as.data.frame(reducedDim(pbmcs[[i]], "UMAP"))
  p <- ggplot2::ggplot() +
    ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                        size = 1, alpha = 1) +
    ggplot2::labs(title = titles[i], x = "UMAP_1", y = "UMAP_2", color = "") +
    ggplot2::theme_classic(base_size = 20) +
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
    ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
  if(i == 1){
    p <- p + ggplot2::scale_colour_hue()
  }else if(i == 2){
    p <- p + ggplot2::scale_colour_brewer(palette = "Set1")
  }else if(i == 3){
    p <- p + ggplot2::scale_colour_brewer(palette = "Set2")
  }
  filename <- sprintf("figures/figure_04_%04d.png", 29 + i)
  ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.1, height = 4.3)
}


6.8 Investigate significant signs

Significant signs are analogous to differentially expressed genes but bear biological meanings. Note that naïve usages of statistical tests should be avoided because the row vectors of SSMs are centered.

Instead, ASURAT function compute_sepI_all() computes separation indices for each cluster against the others. Briefly, a separation index “sepI”, ranging from -1 to 1, is a nonparametric measure of significance of a given sign score for a given subpopulation. The larger (resp. smaller) sepI is, the more reliable the sign is as a positive (resp. negative) marker for the cluster.

for(i in seq_along(pbmcs)){
  set.seed(1)
  labels <- colData(pbmcs[[i]])$seurat_clusters
  pbmcs[[i]] <- compute_sepI_all(sce = pbmcs[[i]], labels = labels,
                                 nrand_samples = NULL)
}

set.seed(1)
pbmcs_LabelCB_SignGO <- pbmcs$GO
metadata(pbmcs_LabelCB_SignGO)$marker_signs <- NULL
lbs <- colData(pbmcs$CB)$seurat_clusters
pbmcs_LabelCB_SignGO <- compute_sepI_all(sce = pbmcs_LabelCB_SignGO,
                                         labels = lbs, nrand_samples = NULL)

set.seed(1)
pbmcs_LabelCB_SignKG <- pbmcs$KG
metadata(pbmcs_LabelCB_SignKG)$marker_signs <- NULL
lbs <- colData(pbmcs$CB)$seurat_clusters
pbmcs_LabelCB_SignKG <- compute_sepI_all(sce = pbmcs_LabelCB_SignKG,
                                         labels = lbs, nrand_samples = NULL)


6.9 Investigate significant genes

6.9.1 Use Seurat function

To date (December, 2021), one of the most useful methods of multiple statistical tests in scRNA-seq data analysis is to use a Seurat function FindAllMarkers().

If there is gene expression data in altExp(sce), one can investigate differentially expressed genes by using Seurat functions in the similar manner as described before.

set.seed(1)
surt <- Seurat::as.Seurat(pbmcs$CB, counts = "counts", data = "counts")
mat <- as.matrix(assay(altExp(pbmcs$CB), "counts"))
surt[["GEM"]] <- Seurat::CreateAssayObject(counts = mat)
Seurat::DefaultAssay(surt) <- "GEM"
surt <- Seurat::SetIdent(surt, value = "seurat_clusters")
res <- Seurat::FindAllMarkers(surt, only.pos = TRUE,
                              min.pct = 0.25, logfc.threshold = 0.25)
metadata(pbmcs$CB)$marker_genes$all <- res
# Save data.
saveRDS(pbmcs, file = "backup/04_005_pbmcs4k_desdeg.rds")
saveRDS(pbmcs_LabelCB_SignGO, file = "backup/04_005_pbmcs4k_LabelCB_SignGO.rds")
saveRDS(pbmcs_LabelCB_SignKG, file = "backup/04_005_pbmcs4k_LabelCB_SignKG.rds")

# Load data.
pbmcs <- readRDS("backup/04_005_pbmcs4k_desdeg.rds")
pbmcs_LabelCB_SignGO <- readRDS("backup/04_005_pbmcs4k_LabelCB_SignGO.rds")
pbmcs_LabelCB_SignKG <- readRDS("backup/04_005_pbmcs4k_LabelCB_SignKG.rds")


6.10 Multifaceted analysis

Simultaneously analyze multiple sign-by-sample matrices, which helps us characterize individual samples (cells) from multiple biological aspects.

ASURAT function plot_multiheatmaps() shows heatmaps (ComplexHeatmap object) of sign scores and gene expression levels (if there are), where rows and columns stand for sign (or gene) and sample (cell), respectively.

First, remove unrelated signs by setting keywords, followed by selecting top significant signs and genes for the clustering results with respect to separation index and p-value, respectively.

# Significant signs
marker_signs <- list()
keys <- "foofoo|hogehoge"
for(i in seq_along(pbmcs)){
  if(i == 1){
    marker_signs[[i]] <- metadata(pbmcs[[i]])$marker_signs$all
  }else if(i == 2){
    marker_signs[[i]] <- metadata(pbmcs_LabelCB_SignGO)$marker_signs$all
  }else if(i == 3){
    marker_signs[[i]] <- metadata(pbmcs_LabelCB_SignKG)$marker_signs$all
  }
  marker_signs[[i]] <- marker_signs[[i]][!grepl(keys, marker_signs[[i]]$Description), ]
  marker_signs[[i]] <- dplyr::group_by(marker_signs[[i]], Ident_1)
  marker_signs[[i]] <- dplyr::slice_max(marker_signs[[i]], sepI, n = 1)
  marker_signs[[i]] <- dplyr::slice_min(marker_signs[[i]], Rank, n = 1)
}
# Significant genes
marker_genes_CB <- metadata(pbmcs$CB)$marker_genes$all
marker_genes_CB <- dplyr::group_by(marker_genes_CB, cluster)
marker_genes_CB <- dplyr::slice_min(marker_genes_CB, p_val_adj, n = 1)
marker_genes_CB <- dplyr::slice_max(marker_genes_CB, avg_log2FC, n = 1)

Then, prepare arguments.

# ssm_list
sces_sub <- list() ; ssm_list <- list()
for(i in seq_along(pbmcs)){
  sces_sub[[i]] <- pbmcs[[i]][rownames(pbmcs[[i]]) %in% marker_signs[[i]]$SignID, ]
  ssm_list[[i]] <- assay(sces_sub[[i]], "counts")
}
names(ssm_list) <- c("SSM_celltype", "SSM_function", "SSM_pathway")
# gem_list
expr_sub <- altExp(pbmcs$CB, "logcounts")
expr_sub <- expr_sub[rownames(expr_sub) %in% marker_genes_CB$gene]
gem_list <- list(x = t(scale(t(as.matrix(assay(expr_sub, "counts"))))))
names(gem_list) <- "Scaled\nLogExpr"
# ssmlabel_list
labels <- list() ; ssmlabel_list <- list()
for(i in seq_along(pbmcs)){
  tmp <- colData(sces_sub[[i]])$seurat_clusters
  labels[[i]] <- data.frame(label = tmp)
  n_groups <- length(unique(tmp))
  if(i == 1){
    labels[[i]]$color <- scales::hue_pal()(n_groups)[tmp]
  }else if(i == 2){
    labels[[i]]$color <- scales::brewer_pal(palette = "Set1")(n_groups)[tmp]
  }else if(i == 3){
    labels[[i]]$color <- scales::brewer_pal(palette = "Set2")(n_groups)[tmp]
  }
  ssmlabel_list[[i]] <- labels[[i]]
}
names(ssmlabel_list) <- c("Label_celltype", "Label_function", "Label_pathway")

Tips: If one would like to omit some color labels (e.g., labels[[3]]), set the argument as follows:

ssmlabel_list[[2]] <- data.frame(label = NA, color = NA)
ssmlabel_list[[3]] <- data.frame(label = NA, color = NA)

Finally, plot heatmaps for the selected signs and genes.

filename <- "figures/figure_04_0040.png"
#png(file = filename, height = 1450, width = 1650, res = 300)
png(file = filename, height = 300, width = 300, res = 60)
set.seed(12)
title <- "PBMC 4k"
plot_multiheatmaps(ssm_list = ssm_list, gem_list = gem_list,
                   ssmlabel_list = ssmlabel_list, gemlabel_list = NULL,
                   nrand_samples = 500, show_row_names = TRUE, title = title)
dev.off()

Show violin plots for the sign score distributions across cell type-related clusters.

labels <- colData(pbmcs$CB)$seurat_clusters
vlist <- list(c("GO", "GO:0050862-S", "...T cell receptor... (LCK, TRAT1, ...)"),
              c("CB", "CL:0000576-S", "monocyte (S100A12, CD74, ...)"),
              c("CB", "MSigDBID:1-V", "...NK_NKT_CELLS... (NKG7, GZMB, ...)"),
              c("CB", "MSigDBID:23-V", "...B_CELLS (BCL11A, CD79A, ...)"),
              c("CB", "CellMarkerID:72-S", "Dendritic cell (CD83, CD302, ...)"))
for(i in seq_along(vlist)){
  ind <- which(rownames(pbmcs[[vlist[[i]][1]]]) == vlist[[i]][2])
  subsce <- pbmcs[[vlist[[i]][1]]][ind, ]
  df <- as.data.frame(t(as.matrix(assay(subsce, "counts"))))
  p <- ggplot2::ggplot() +
    ggplot2::geom_violin(ggplot2::aes(x = as.factor(labels), y = df[, 1],
                                      fill = labels), trim = FALSE, size = 0.5) +
    ggplot2::geom_boxplot(ggplot2::aes(x = as.factor(labels), y = df[, 1]),
                          width = 0.15, alpha = 0.6) +
    ggplot2::labs(title = paste0(vlist[[i]][2], "\n", vlist[[i]][3]),
                  x = "Cluster (cell type)", y = "Sign score", fill = "Cluster") +
    ggplot2::theme_classic(base_size = 25) +
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 20),
                   legend.position = "none") +
    ggplot2::scale_fill_hue()
  filename <- sprintf("figures/figure_04_%04d.png", 49 + i)
  ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 6, height = 3.5)
}


6.11 Infer cell types

colData(pbmcs$CB)$cell_type <- as.character(colData(pbmcs$CB)$seurat_clusters)
colData(pbmcs$CB)$cell_type[colData(pbmcs$CB)$cell_type == 0] <- "T"
colData(pbmcs$CB)$cell_type[colData(pbmcs$CB)$cell_type == 1] <- "Mono"
colData(pbmcs$CB)$cell_type[colData(pbmcs$CB)$cell_type == 2] <- "NK/NKT"
colData(pbmcs$CB)$cell_type[colData(pbmcs$CB)$cell_type == 3] <- "B"
colData(pbmcs$CB)$cell_type[colData(pbmcs$CB)$cell_type == 4] <- "DC"

Show the annotation results in low-dimensional spaces.

title <- "PBMC 4k (cell type)"
labels <- factor(colData(pbmcs$CB)$cell_type,
                 levels = c("T", "Mono", "NK/NKT", "B", "DC"))
df <- as.data.frame(reducedDim(pbmcs$CB, "UMAP"))
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "Cell state") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::scale_colour_hue() +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
filename <- "figures/figure_04_0080.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.8, height = 4.3)

# Save data.
saveRDS(pbmcs, file = "backup/04_006_pbmcs4k_annotation.rds")

# Load data.
pbmcs <- readRDS("backup/04_006_pbmcs4k_annotation.rds")


7 Using the existing softwares

7.1 scran

Load the normalized data (see here).

pbmc <- readRDS("backup/04_003_pbmc4k_normalized.rds")

Prepare a SingleCellExperiment object.

mat <- as.matrix(assay(pbmc, "counts"))
pbmc <- SingleCellExperiment(assays = list(counts = as.matrix(mat)),
                             rowData = data.frame(gene = rownames(mat)),
                             colData = data.frame(sample = colnames(mat)))


7.1.1 Normalize data

According to the scran protocol, perform cell clustering for computing size factors for individual clusters, normalize data within individual clusters, and perform a variance modeling for each gene.

# Quick cell clustering
clusters <- scran::quickCluster(pbmc)
# scran normalization
pbmc <- scran::computeSumFactors(pbmc, clusters = clusters)
pbmc <- scater::logNormCounts(pbmc)
# Perform a variance modeling
metadata(pbmc)$dec <- scran::modelGeneVar(pbmc)

Show the results of variance modeling.

df <- data.frame(x = metadata(pbmc)$dec$mean, y = metadata(pbmc)$dec$total,
                 z = metadata(pbmc)$dec$tech)
title <- "PBMC 4k"
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[,1], y = df[,2]), color = "black",
                      alpha = 1, size = 2) +
  ggplot2::geom_line(ggplot2::aes(x = df[,1], y = df[,3]), color = "red",
                     alpha = 1, size = 2) +
  ggplot2::labs(title = title, x = "Mean log-expression", y = "Variance") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 20))
filename <- "figures/figure_04_0110.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 4.0, height = 4.0)


7.1.2 Cluster cells

According to the scran protocol, choose top variable genes using getTopHVGs(), performe denoisPCA() by inputting the variable genes, and perform k-nearest neighbor graph-based clustering, where k is a resolution parameter.

Tips: As mentioned in Cruz and Wishart, Cancer Inform. 2, 59-77 (2006), highly variable genes are selected as the cell-per-variable gene ratio being 5:1.

# Choose top variable genes.
hvg <- scran::getTopHVGs(metadata(pbmc)$dec, n = round(0.2 * ncol(pbmc)))
# Principal component analysis
pbmc <- scran::denoisePCA(pbmc, metadata(pbmc)$dec, subset.row = hvg)
# Cell clustering
set.seed(1)
g <- scran::buildSNNGraph(pbmc, use.dimred = "PCA", k = 200, type = "rank")
c <- igraph::cluster_louvain(g)$membership
colData(pbmc)$scran_clusters <- as.factor(c)


7.1.3 Reduce dimensions

Perform t-distributed stochastic neighbor embedding and Uniform Manifold Approximation and Projection.

mat <- reducedDim(pbmc, "PCA")
# t-SNE
set.seed(1)
res <- Rtsne::Rtsne(mat, dim = 2, pca = FALSE)
reducedDim(pbmc, "TSNE") <- res[["Y"]]
# UMAP
set.seed(1)
res <- umap::umap(mat, n_components = 2)
reducedDim(pbmc, "UMAP") <- res[["layout"]]

Show the clustering results.

title <- "PBMC 4k (scran)"
labels <- colData(pbmc)$scran_clusters
df <- as.data.frame(reducedDim(pbmc, "UMAP"))
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 20)) +
  ggplot2::scale_colour_hue() +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
filename <- "figures/figure_04_0130.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.1, height = 4.3)


7.1.4 Find differentially expressed genes

Performs pairwiseTTests() and combineMarkers() for finding cluster markers.

metadata(pbmc)$pwtt <- scran::pairwiseTTests(
  x = as.matrix(assay(pbmc, "logcounts")),
  groups = colData(pbmc)$scran_clusters, direction = "up")
metadata(pbmc)$cmb <- scran::combineMarkers(
  de.lists = metadata(pbmc)$pwtt$statistics,
  pairs = metadata(pbmc)$pwtt$pairs, pval.type = "all")
# Create a result table.
res <- list()
for(i in seq_along(metadata(pbmc)$cmb@listData)){
  g <- metadata(pbmc)$cmb@listData[[i]]@rownames
  p <- metadata(pbmc)$cmb@listData[[i]]@listData[["p.value"]]
  fd <- metadata(pbmc)$cmb@listData[[i]]@listData[["FDR"]]
  fc <- metadata(pbmc)$cmb@listData[[i]]@listData[["summary.logFC"]]
  res[[i]] <- data.frame(label = i, gene = g, pval = p, FDR = fd, logFC = fc)
}
tmp <- c()
for(i in seq_along(res)){
  tmp <- rbind(tmp, res[[i]])
}
metadata(pbmc)$stat <- tmp
View(metadata(pbmc)$stat[metadata(pbmc)$stat$FDR < 10^(-100), ])


7.1.5 Infer cell types

Defining significant genes as genes with FDR<1e-100, infer cell types using GeneCards.

1: Monocyte     # S100A9 (FDR ~0), MNDA (FDR ~0)
2: NK/NKT       # NKG7 (FDR ~e-217), GZMA (FDR ~e-179)
3: Unspecified  # Only JUNB was detected as significant genes.
4: B cell       # CD79A (FDR ~e-311), CD79B (FDR ~e-283)
5: Unspecified  # No significant genes were detected.
colData(pbmc)$cell_type <- as.integer(as.character(colData(pbmc)$scran_clusters))
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 1] <- "Mono"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 2] <- "NK/NKT"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 3] <- "Unspecified"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 4] <- "B"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 5] <- "Unspecified"

Show the annotation results in low-dimensional spaces.

title <- "PBMC 4k (scran)"
labels <- factor(colData(pbmc)$cell_type,
                 levels = c("Mono", "NK/NKT", "B", "Unspecified"))
df <- as.data.frame(reducedDim(pbmc, "UMAP"))
mycolor <- scales::hue_pal()(5)
mycolor <- c("Mono" = mycolor[2], "NK/NKT" = mycolor[3], "B" = mycolor[4],
             "Unspecified" = "grey80")
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "Cell type") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4))) +
  ggplot2::scale_color_manual(values = mycolor)
filename <- "figures/figure_04_0140.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 6.1, height = 4.3)

# Save data.
saveRDS(pbmc, file = "backup/04_011_pbmc4k_scran.rds")

# Load data.
pbmc <- readRDS("backup/04_011_pbmc4k_scran.rds")


7.2 Seurat

Load the normalized data (see here).

pbmc <- readRDS("backup/04_003_pbmc4k_normalized.rds")

Create Seurat objects.

pbmc <- Seurat::CreateSeuratObject(counts = as.matrix(assay(pbmc, "counts")),
                                   project = "PBMC")


7.2.1 Perform Seurat preprocessing

According to the Seurat protocol, normalize data, perform variance stabilizing transform by setting the number of variable feature, scale data, and reduce dimension using principal component analysis.

# Normalization
pbmc <- Seurat::NormalizeData(pbmc, normalization.method = "LogNormalize")
# Variance stabilizing transform
n <- round(0.9 * ncol(pbmc))
pbmc <- Seurat::FindVariableFeatures(pbmc, selection.method = "vst", nfeatures = n)
# Scale data
pbmc <- Seurat::ScaleData(pbmc)
# Principal component analysis
pbmc <- Seurat::RunPCA(pbmc, features = Seurat::VariableFeatures(pbmc))


7.2.2 Cluster cells

Compute the cumulative sum of variances, which is used for determining the number of the principal components (PCs).

pc <- which(cumsum(pbmc@reductions[["pca"]]@stdev) /
              sum(pbmc@reductions[["pca"]]@stdev) > 0.9)[1]

Perform cell clustering.

# Create k-nearest neighbor graph.
pbmc <- Seurat::FindNeighbors(pbmc, reduction = "pca", dim = seq_len(pc))
# Cluster cells.
pbmc <- Seurat::FindClusters(pbmc, resolution = 0.08)
# Run t-SNE.
pbmc <- Seurat::RunTSNE(pbmc, dims.use = seq_len(2), reduction = "pca",
                        dims = seq_len(pc), do.fast = FALSE, perplexity = 30)
# Run UMAP.
pbmc <- Seurat::RunUMAP(pbmc, dims = seq_len(pc))

Show the clustering results.

title <- "PBMC 4k (Seurat)"
labels <- pbmc$seurat_clusters
df <- pbmc@reductions[["umap"]]@cell.embeddings
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::scale_colour_hue() +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
filename <- "figures/figure_04_0230.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.1, height = 4.3)


7.2.3 Find differentially expressed genes

Find differentially expressed genes.

pbmc@misc$stat <- Seurat::FindAllMarkers(pbmc, only.pos = TRUE, min.pct = 0.25,
                                         logfc.threshold = 0.25)
View(pbmc@misc$stat[which(pbmc@misc$stat$p_val_adj < 10^(-100)), ])


7.2.4 Infer cell types

Defining significant genes as genes with FDR<1e-100, infer cell types using GeneCards.

0: T cell          # TRAC (p_val_adj ~0), CD3E (p_val_adj ~e-273)
1: Monocyte        # S100A8 (p_val_adj ~0), S100A9 (p_val_adj ~0)
2: NK/NKT          # NKG7 (p_val_adj ~0), GZMA (p_val_adj ~0)
3: B cell          # CD79A (p_val_adj ~0), CD79B (p_val_adj ~0)
4: Dendritic cell  # LILRA4 (p_val_adj ~e-270)
tmp <- as.integer(as.character(pbmc$seurat_clusters))
pbmc$cell_type <- tmp
pbmc$cell_type[pbmc$cell_type == 0] <- "T"
pbmc$cell_type[pbmc$cell_type == 1] <- "Mono"
pbmc$cell_type[pbmc$cell_type == 2] <- "NK/NKT"
pbmc$cell_type[pbmc$cell_type == 3] <- "B"
pbmc$cell_type[pbmc$cell_type == 4] <- "DC"

Show the annotation results in low-dimensional spaces.

title <- "PBMC 4k (Seurat)"
labels <- factor(pbmc$cell_type, levels = c("T", "Mono", "NK/NKT", "B", "DC"))
df <- pbmc@reductions[["umap"]]@cell.embeddings
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "Cell type") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::scale_colour_hue() +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
filename <- "figures/figure_04_0240.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.8, height = 4.3)

Show gene expression levels in low-dimensional spaces.

title <- "PBMC 4k (Seurat)"
genes <- c("TRAC", "LYZ", "NKG7", "CD79A", "LILRA4")
labels <- pbmc$seurat_clusters
mat <- as.matrix(Seurat::GetAssayData(object = pbmc, slot = "data"))
for(i in seq_along(genes)){
  expr <- mat[which(rownames(mat) == genes[i]), ]
  p <- ggplot2::ggplot() +
    ggplot2::geom_violin(ggplot2::aes(x = as.factor(labels), y = expr,
                                      fill = labels), trim = FALSE, size = 0.5) +
    ggplot2::geom_boxplot(ggplot2::aes(x = as.factor(labels), y = expr),
                          width = 0.15, alpha = 0.6) +
    ggplot2::labs(title = genes[i], x = "Cluster (Seurat)", y = "Expression",
                  fill = "Cluster") +
    ggplot2::theme_classic(base_size = 25) +
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 25),
                   legend.position = "none") +
    ggplot2::scale_fill_hue()
  filename <- sprintf("figures/figure_04_0241_%02d.png", i)
  ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 4, height = 3.4)
}

# Save data.
saveRDS(pbmc, file = "backup/04_021_pbmc4k_seurat.rds")

# Load data.
pbmc <- readRDS("backup/04_021_pbmc4k_seurat.rds")


7.2.5 Infer cell types using Seurat and scCATCH

Check the package version.

packageVersion("scCATCH")
[1] ‘3.0’

Load Seurat computational results.

pbmc <- readRDS("backup/04_021_pbmc4k_seurat.rds")

Create scCATCH objects.

mat <- as.matrix(Seurat::GetAssayData(object = pbmc, slot = "data"))
mat <- scCATCH::rev_gene(data = mat, data_type = "data", species = "Human",
                         geneinfo = scCATCH::geneinfo)
labels <- as.character(pbmc$seurat_clusters)
scc <- scCATCH::createscCATCH(data = mat, cluster = labels)

Perform findmarkergenes() and findcelltype().

scc <- scCATCH::findmarkergene(object = scc, species = "Human",
                               marker = scCATCH::cellmatch,
                               tissue = "Peripheral blood", cancer = "Normal",
                               cell_min_pct = 0.25, logfc = 0.25, pvalue = 0.05)
scc <- scCATCH::findcelltype(object = scc)
pbmc@misc[["scCATCH"]] <- scc
View(pbmc@misc[["scCATCH"]]@celltype)

Annotate cells.

tmp <- pbmc[[]]
tmp[which(tmp$seurat_clusters == 0), ]$cell_type <- "Unspecified"
tmp[which(tmp$seurat_clusters == 1), ]$cell_type <- "Mono"
tmp[which(tmp$seurat_clusters == 2), ]$cell_type <- "NK/NKT"
tmp[which(tmp$seurat_clusters == 3), ]$cell_type <- "B"
tmp[which(tmp$seurat_clusters == 4), ]$cell_type <- "T"
pbmc <- Seurat::AddMetaData(pbmc, tmp)

Show the annotation results in low-dimensional spaces.

title <- "PBMC 4k (Seurat + scCatch)"
labels <- factor(pbmc[[]]$cell_type,
                 levels = c("T", "Mono", "B", "NK/NKT", "Unspecified"))
mycolor <- scales::hue_pal()(5)
mycolor <- c("T" = mycolor[1], "Mono" = mycolor[2], "NK/NKT" = mycolor[3],
             "B" = mycolor[4], "Unspecified" = "grey80")
df <- pbmc@reductions[["umap"]]@cell.embeddings
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "Cell type") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::scale_color_manual(values = mycolor) +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
filename <- "figures/figure_04_0245.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 6.1, height = 4.3)

# Save data.
saveRDS(pbmc, file = "backup/04_022_pbmc4k_seurat_sccatch.rds")

# Load data.
pbmc <- readRDS("backup/04_022_pbmc4k_seurat_sccatch.rds")


7.2.6 Infer cell types using ssGSEA

Load the Seurat annotation results.

pbmc <- readRDS("backup/04_021_pbmc4k_seurat.rds")

Check the package version.

packageVersion("escape")
[1] ‘1.0.1’

Perform getGeneSets() with an argument library = "C8" (“cell type signature gene sets” in MSigDB).

pbmc@misc[["getGeneSets"]] <- escape::getGeneSets(species = "Homo sapiens",
                                                  library = "C8")

Perform enrichIt(), estimating ssGSEA scores, in which the arguments are the same with those in the vignettes in escape package.

ES <- escape::enrichIt(obj = pbmc, gene.sets = pbmc@misc[["getGeneSets"]],
                       groups = 1000, cores = 4)
pbmc <- Seurat::AddMetaData(pbmc, ES)
pbmc@misc[["enrichIt"]] <- ES
[1] "Using sets of 1000 cells. Running 5 times."
Setting parallel calculations through a SnowParam back-end
with workers=4 and tasks=100.
Estimating ssGSEA scores for 654 gene sets.
Setting parallel calculations through a SnowParam back-end
with workers=4 and tasks=100.
Estimating ssGSEA scores for 654 gene sets.
...
Setting parallel calculations through a SnowParam back-end
with workers=4 and tasks=100.
Estimating ssGSEA scores for 654 gene sets.

Perform t-distributed stochastic neighbor embedding and Uniform Manifold Approximation and Projection.

mat <- pbmc@misc[["enrichIt"]]
# t-SNE
set.seed(1)
res <- Rtsne::Rtsne(mat, dim = 2, pca = TRUE, initial_dims = 50)
pbmc@reductions[["tsne_ssgsea"]] <- res[["Y"]]
# UMAP
set.seed(1)
res <- umap::umap(mat, n_components = 2)
pbmc@reductions[["umap_ssgsea"]] <- res[["layout"]]

Perform unsupervised clustering of cells using Seurat functions.

surt <- Seurat::CreateSeuratObject(counts = t(pbmc@misc[["enrichIt"]]),
                                   project = "PBMC")
surt <- Seurat::ScaleData(surt, features = rownames(surt))
surt <- Seurat::RunPCA(surt, features = rownames(surt))
surt <- Seurat::FindNeighbors(surt, reduction = "pca", dims = seq_len(40))
surt <- Seurat::FindClusters(surt, resolution = 0.1)
pbmc <- Seurat::AddMetaData(pbmc, metadata = surt[[]]$seurat_clusters,
                            col.name = "ssgsea_clusters")

Show the clustering results in low-dimensional spaces.

labels <- pbmc$ssgsea_clusters
df <- as.data.frame(pbmc@reductions[["umap_ssgsea"]])
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 0.2, alpha = 1) +
  ggplot2::labs(title = "PBMC 4k (ssGSEA)", x = "UMAP_1", y = "UMAP_2",
                color = "") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4),
                                                 ncol = 1))
filename <- "figures/figure_04_0280.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.1, height = 4.3)

# Save data.
saveRDS(pbmc, file = "backup/04_023_pbmc4k_ssGSEA_msigdb.rds")

# Load data.
pbmc <- readRDS("backup/04_023_pbmc4k_ssGSEA_msigdb.rds")


7.2.7 Infer cell types using Seurat and ssGSEA

Load ssGSEA results.

pbmc <- readRDS("backup/04_023_pbmc4k_ssGSEA_msigdb.rds")

Investigate “significant” modules for Seurat clustering results.

set.seed(1)
surt <- Seurat::CreateSeuratObject(counts = t(pbmc@misc[["enrichIt"]]))
surt <- Seurat::AddMetaData(surt, metadata = pbmc$cell_type,
                            col.name = "seurat_cell_type")
surt <- Seurat::SetIdent(surt, value = "seurat_cell_type")
res <- Seurat::FindAllMarkers(surt, only.pos = TRUE,
                              min.pct = 0.15, logfc.threshold = 0.15)
rownames(res) <- seq_len(nrow(res))
pbmc@misc[["ssGSEA_stat"]] <- res

Show violin plots for ssGSEA score distributions for Seurat annotation results.

lvs <- c("T", "Mono", "NK/NKT", "B", "DC")
labels <- factor(pbmc$cell_type, levels = lvs)
descriptions <- c(gsub("-", "_", "DURANTE-ADULT-OLFACTORY-NEUROEPITHELIUM-CD4-T-CELLS"),
                  gsub("-", "_", "TRAVAGLINI-LUNG-CLASSICAL-MONOCYTE-CELL"),
                  gsub("-", "_", "TRAVAGLINI-LUNG-NATURAL-KILLER-T-CELL"),
                  gsub("-", "_", "DURANTE-ADULT-OLFACTORY-NEUROEPITHELIUM-B-CELLS"),
                  gsub("-", "_", "TRAVAGLINI-LUNG-PLASMACYTOID-DENDRITIC-CELL"),
                  gsub("-", "_", "TRAVAGLINI-LUNG-MYELOID-DENDRITIC-TYPE-1-CELL"))
titles <- c(paste0("MSigDB: ssGSEA score for\n", gsub("_", "-", descriptions[1])),
            paste0("MSigDB: ssGSEA score for\n", gsub("_", "-", descriptions[2])),
            paste0("MSigDB: ssGSEA score for\n", gsub("_", "-", descriptions[3])),
            paste0("MSigDB: ssGSEA score for\n", gsub("_", "-", descriptions[4])),
            paste0("MSigDB: ssGSEA score for\n", gsub("_", "-", descriptions[5])),
            paste0("MSigDB: ssGSEA score for\n", gsub("_", "-", descriptions[6])))
mat <- pbmc@misc[["enrichIt"]]
mat <- mat[, which(colnames(mat) %in% descriptions)]
for(i in seq_along(descriptions)){
  df <- data.frame(label = labels,
                   ssGSEA = mat[, which(colnames(mat) == descriptions[i])])
#  df <- tidyr::pivot_longer(df, cols = c("Sign", "ssGSEA"))
  p <- ggplot2::ggplot() +
    ggplot2::geom_violin(ggplot2::aes(x = as.factor(df$label), y = df$ssGSEA),
                         fill = "grey80", trim = FALSE, size = 1) +
    ggplot2::labs(title = titles[i], x = "Seurat annotations",
                  y = "ssGSEA scores") +
    ggplot2::scale_x_discrete(guide = ggplot2::guide_axis(angle = 45)) +
    ggplot2::theme_classic(base_size = 20) +
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 15))
  filename <- sprintf("figures/figure_04_%04d.png", 289 + i)
  ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 6, height = 4.5)
}

# Save data.
saveRDS(pbmc, file = "backup/04_024_pbmc4k_seurat_ssGSEA.rds")

# Load data.
pbmc <- readRDS("backup/04_024_pbmc4k_seurat_ssGSEA.rds")


7.3 Monocle 3

Load the normalized data (see here).

pbmc <- readRDS("backup/04_003_pbmc4k_normalized.rds")

Create Monocle 3 objects (cell_data_set (CDS)).

gene_metadata <- data.frame(gene_short_name = rowData(pbmc)$gene)
rownames(gene_metadata) <- rowData(pbmc)$gene
pbmc <- monocle3::new_cell_data_set(
  expression_data = as.matrix(assay(pbmc, "counts")),
  cell_metadata = colData(pbmc),
  gene_metadata = gene_metadata)


7.3.1 Perform Monocle 3 preprocessing

Preprocess the data under the default settings of Monocle 3 (version 1.0.0), e.g., num_dim = 50 and norm_method = "log".

pbmc <- monocle3::preprocess_cds(pbmc)

Perform log-normalization with a pseudo count.

assay(pbmc, "logcounts") <- monocle3::normalized_counts(pbmc,
                                                        norm_method = "log",
                                                        pseudocount = 1)

Perform dimensionality reduction.

#pbmc <- monocle3::reduce_dimension(pbmc, reduction_method = "tSNE",
#                                   preprocess_method = "PCA")
pbmc <- monocle3::reduce_dimension(pbmc, reduction_method = "UMAP",
                                   preprocess_method = "PCA")


7.3.2 Cluster cells

Group cells into several clusters by using a community detection algorithm.

#pbmc <- monocle3::cluster_cells(pbmc, resolution = 5e-4, reduction_method = "tSNE")
pbmc <- monocle3::cluster_cells(pbmc, resolution = 5e-5, reduction_method = "UMAP")

Show the clustering results.

title <- "PBMC 4k (Monocle 3)"
labels <- pbmc@clusters@listData[["UMAP"]][["clusters"]]
df <- reducedDim(pbmc, "UMAP")
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::scale_colour_hue() +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4)))
filename <- "figures/figure_04_0330.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5.1, height = 4.3)


7.3.3 Find differentially expressed genes

Find differentially expressed genes.

set.seed(1)
metadata(pbmc)$markers <- monocle3::top_markers(pbmc, group_cells_by = "cluster",
                                                reference_cells = 1000, cores = 2)
markers <- metadata(pbmc)$markers
markers <- markers[order(markers$cell_group, markers$marker_test_q_value), ]
View(markers[which(markers$marker_test_q_value < 10^(-100)), ])


7.3.4 Infer cell types

Defining significant genes as genes with marker_teset_q_value<1e-100, infer cell types using GeneCards.

1: T cell          # TRAC (marker_teset_q_value ~e-268)
                   # CD3D (marker_teset_q_value ~e-210)
2: Monocyte        # S100A9 (marker_teset_q_value ~e-300)
                   # S100A8 (marker_teset_q_value ~e-282)
3: B cell          # CD79A (marker_teset_q_value ~0)
                   # CD79B (marker_teset_q_value ~e-305)
4: NK/NKT          # GZMK (marker_teset_q_value ~e-134)
                   # NKG7 (marker_teset_q_value ~e-105)
5: NK/NKT          # GNLY (marker_teset_q_value ~e-132)
                   # NKG7 (marker_teset_q_value ~e-125)
6: Unspecified     # No significant genes are detected.
                   # No significant genes are detected.
7: Unspecified     # No significant genes are detected.
8: Unspecified     # No significant genes are detected.
tmp <- as.integer(as.character(pbmc@clusters@listData[["UMAP"]][["clusters"]]))
colData(pbmc)$cell_type <- tmp
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 1] <- "T"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 2] <- "Mono"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 3] <- "B"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 4] <- "NK/NKT"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 5] <- "NK/NKT"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 6] <- "Unspecified"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 7] <- "Unspecified"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 8] <- "Unspecified"

Show the annotation results in low-dimensional spaces.

title <- "PBMC 4k (Monocle 3)"
labels <- factor(colData(pbmc)$cell_type,
                 levels = c("T", "Mono", "B", "NK/NKT", "Unspecified"))
mycolor <- scales::hue_pal()(5)
mycolor <- c("T" = mycolor[1], "Mono" = mycolor[2], "NK/NKT" = mycolor[3],
             "B" = mycolor[4], "Unspecified" = "grey80")
df <- reducedDim(pbmc, "UMAP")
p <- ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(x = df[, 1], y = df[, 2], color = labels),
                      size = 1, alpha = 1) +
  ggplot2::labs(title = title, x = "UMAP_1", y = "UMAP_2", color = "Cell type") +
  ggplot2::theme_classic(base_size = 20) +
  ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18)) +
  ggplot2::guides(colour = ggplot2::guide_legend(override.aes = list(size = 4))) +
  ggplot2::scale_color_manual(values = mycolor)
filename <- "figures/figure_04_0340.png"
ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 6.2, height = 4.3)

# Save data.
saveRDS(pbmc, file = "backup/04_031_pbmc4k_monocle3.rds")

# Load data.
pbmc <- readRDS("backup/04_031_pbmc4k_monocle3.rds")


7.4 SC3

Load the normalized data (see here).

pbmc <- readRDS("backup/04_003_pbmc4k_normalized.rds")

Prepare a SingleCellExperiment object.

mat <- as.matrix(assay(pbmc, "counts"))
pbmc <- SingleCellExperiment(assays = list(counts = as.matrix(mat),
                                           logcounts = log2(mat + 1)),
                             rowData = data.frame(gene = rownames(mat)),
                             colData = data.frame(sample = colnames(mat)))
rowData(pbmc)$feature_symbol <- rownames(pbmc)


7.5 Cluster cells

Run sc3(), in which parameter ks is subjectively determined by considering biological backgrounds. Here, sc3() could not compute for sc68_vehi and sc68_cisp.

set.seed(1)
pbmc <- SC3::sc3(pbmc, ks = 4:8, biology = TRUE)

Plot stability indices across ks for investigating “optimal” number of clusters.

kss <- as.integer(names(metadata(pbmc)$sc3$consensus))
for(i in seq_along(kss)){
  p <- SC3::sc3_plot_cluster_stability(pbmc, k = kss[i])
  p <- p + ggplot2::labs(title = paste0("PBMC 4k (SC3) k = ", kss[i])) +
    ggplot2::theme_classic(base_size = 20) +
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 18))
  filename <- sprintf("figures/figure_04_%04d.png", 409 + i)
  ggplot2::ggsave(file = filename, plot = p, dpi = 50, width = 5, height = 4)
}


7.5.1 Find differentially expressed genes

Setting the optimal number of clusters, find differentially expressed genes.

row_data <- rowData(pbmc)
markers <- as.data.frame(row_data[ , grep("gene|sc3_8", colnames(row_data))])
markers <- markers[order(markers$sc3_8_markers_clusts, markers$sc3_8_de_padj), ]
View(markers[which(markers$sc3_8_de_padj < 10^(-100)), ])


7.5.2 Infer cell types

Defining significant genes as genes with sc3_8_de_padj<1e-100, infer cell types using GeneCards.

1: Monocyte        # S100A9 (sc3_8_de_padj ~0)
                   # S100A8 (sc3_8_de_padj ~0)
1: T cell          # LEF1 (sc3_8_de_padj ~e-309)
                   # CCR7 (sc3_8_de_padj ~e-244)
3: B cell          # BANK1 (sc3_8_de_padj ~0)
                   # MS4A1 (sc3_8_de_padj ~0)
4: NK/NKT cell     # NKG7 (sc3_8_de_padj ~0)
                   # KLRD1 (sc3_8_de_padj ~0)
5: T cell          # IL1B (sc3_8_de_padj ~0)
                   # CD69 (sc3_8_de_padj ~e-189)
6: T cell          # LCK (sc3_8_de_padj ~0)
                   # CD27 (sc3_8_de_padj ~e-263)
                   # TCF7 (sc3_8_de_padj ~e-211)
7: NK/NKT cell     # GZMA (sc3_8_de_padj ~0)
                   # TRGC2 (sc3_8_de_padj ~e-316)
                   # CD8A (sc3_8_de_padj ~e-165)
8: T cell          # CD3E (sc3_8_de_padj ~0)
                   # CD3G (sc3_8_de_padj ~0)
                   # TRAC (sc3_8_de_padj ~0)
col_data <- colData(pbmc)
clusters <- col_data[ , grep("sc3_8_clusters", colnames(col_data))]
colData(pbmc)$cell_type <- as.integer(as.character(clusters))
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 1] <- "Mono"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 2] <- "T"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 3] <- "B"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 4] <- "NK/NKT"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 5] <- "T"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 6] <- "T"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 7] <- "NK/NKT"
colData(pbmc)$cell_type[colData(pbmc)$cell_type == 8] <- "T"
# Save data.
saveRDS(pbmc, file = "backup/04_041_pbmc4k_sc3.rds")

# Load data.
pbmc <- readRDS("backup/04_041_pbmc4k_sc3.rds")