Visualization of gene expression with Nebulosa (in Seurat)

Overview

Due to the sparsity observed in single-cell data (e.g. RNA-seq, ATAC-seq), the visualization of cell features (e.g. gene, peak) is frequently affected and unclear, especially when it is overlaid with clustering to annotate cell types. Nebulosa is an R package to visualize data from single cells based on kernel density estimation. It aims to recover the signal from dropped-out features by incorporating the similarity between cells allowing a “convolution” of the cell features.

Import libraries

For this vignette, let’s use Nebulosa with the Seurat package. First, we’ll do a brief/standard data processing.

library("Nebulosa")
library("Seurat")
## Loading required package: SeuratObject
## Loading required package: sp
## 
## Attaching package: 'sp'
## The following object is masked from 'package:IRanges':
## 
##     %over%
## 'SeuratObject' was built under R 4.4.0 but the current version is
## 4.4.1; it is recomended that you reinstall 'SeuratObject' as the ABI
## for R may have changed
## 'SeuratObject' was built with package 'Matrix' 1.7.0 but the current
## version is 1.7.1; it is recomended that you reinstall 'SeuratObject' as
## the ABI for 'Matrix' may have changed
## 
## Attaching package: 'SeuratObject'
## The following object is masked from 'package:SummarizedExperiment':
## 
##     Assays
## The following object is masked from 'package:GenomicRanges':
## 
##     intersect
## The following object is masked from 'package:GenomeInfoDb':
## 
##     intersect
## The following object is masked from 'package:IRanges':
## 
##     intersect
## The following object is masked from 'package:S4Vectors':
## 
##     intersect
## The following object is masked from 'package:BiocGenerics':
## 
##     intersect
## The following objects are masked from 'package:base':
## 
##     intersect, t
## 
## Attaching package: 'Seurat'
## The following object is masked from 'package:SummarizedExperiment':
## 
##     Assays
library("BiocFileCache")

Data pre-processing

Let’s download a dataset of 3k PBMCs (available from 10X Genomics). This same dataset is commonly used in Seurat vignettes. The code below will download, store, and uncompress the data in a temporary directory.

bfc <- BiocFileCache(ask = FALSE)
data_file <- bfcrpath(bfc, file.path(
  "https://s3-us-west-2.amazonaws.com/10x.files/samples/cell",
  "pbmc3k",
  "pbmc3k_filtered_gene_bc_matrices.tar.gz"
))

untar(data_file, exdir = tempdir())

Then, we can read the gene expression matrix using the Read10X from Seurat

data <- Read10X(data.dir = file.path(tempdir(),
  "filtered_gene_bc_matrices",
  "hg19"
))

Let’s create a Seurat object with features being expressed in at least 3 cells and cells expressing at least 200 genes.

pbmc <- CreateSeuratObject(
  counts = data,
  project = "pbmc3k",
  min.cells = 3,
  min.features = 200
)
## Warning: Feature names cannot have underscores ('_'), replacing with dashes
## ('-')

Remove outlier cells based on the number of genes being expressed in each cell (below 2500 genes) and expression of mitochondrial genes (below 5%).

pbmc[["percent.mt"]] <- PercentageFeatureSet(pbmc, pattern = "^MT-")
pbmc <- subset(pbmc, subset = nFeature_RNA < 2500 & percent.mt < 5)

Data normalization

Let’s use SCTransform to stabilize the variance of the data by regressing out the effect of the sequencing depth from each cell.

pbmc <- SCTransform(pbmc, verbose = FALSE)

Dimensionality reduction

Once the data is normalized and scaled, we can run a Principal Component Analysis (PCA) first to reduce the dimensions of our data from 26286 features to 50 principal components. To visualize the principal components, we can run a Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) using the first 30 principal components to obtain a two-dimentional space.

pbmc <- RunPCA(pbmc)
pbmc <- RunUMAP(pbmc, dims = 1:30)

Clustering

To assess cell similarity, let’s cluster the data by constructing a Shared Nearest Neighbor (SNN) Graph using the first 30 principal components and applying the Louvain algorithm.

pbmc <- FindNeighbors(pbmc, dims = 1:30)
pbmc <- FindClusters(pbmc)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
## 
## Number of nodes: 2638
## Number of edges: 108369
## 
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8426
## Number of communities: 13
## Elapsed time: 0 seconds

Visualize data with Nebulosa

The main function from Nebulosa is the plot_density. For usability, it resembles the FeaturePlot function from Seurat.

Let’s plot the kernel density estimate for CD4 as follows

plot_density(pbmc, "CD4")

For comparison, let’s also plot a standard scatterplot using Seurat

FeaturePlot(pbmc, "CD4")

FeaturePlot(pbmc, "CD4", order = TRUE)

By smoothing the data, Nebulosa allows a better visualization of the global expression of CD4 in myeloid and CD4+ T cells. Notice that the “random” expression of CD4 in other areas of the plot is removed as the expression of this gene is not supported by many cells in those areas. Furthermore, CD4+ cells appear to show considerable dropout rate.

Let’s plot the expression of CD4 with Nebulosa next to the clustering results

DimPlot(pbmc, label = TRUE, repel = TRUE)

We can now easily identify that clusters 0 and 2 correspond to CD4+ T cells if we plot CD3D too.

plot_density(pbmc, "CD3D")

Multi-feature visualization

Characterize cell populations usually relies in more than a single marker. Nebulosa allows the visualization of the joint density of from multiple features in a single plot.

Identifying Naive CD8+ T cells

Users familiarized with PBMC datasets may know that CD8+ CCR7+ cells usually cluster next to CD4+ CCR7+ and separate from the rest of CD8+ cells. Let’s aim to identify Naive CD8+ T cells. To do so, we can just add another gene to the vector containing the features to visualize.

p3 <- plot_density(pbmc, c("CD8A", "CCR7"))
p3 + plot_layout(ncol = 1)

Nebulosa can return a joint density plot by multiplying the densities from all query genes by using the joint = TRUE parameter:

p4 <- plot_density(pbmc, c("CD8A", "CCR7"), joint = TRUE)
p4 + plot_layout(ncol = 1)

When compared to the clustering results, we can easily identify that Naive CD8+ T cells correspond to cluster 8.

Nebulosa returns the density estimates for each gene along with the joint density across all provided genes. By setting combine = FALSE, we can obtain a list of ggplot objects where the last plot corresponds to the joint density estimate.

p_list <- plot_density(pbmc, c("CD8A", "CCR7"), joint = TRUE, combine = FALSE)
p_list[[length(p_list)]]

Identifying Naive CD4+ T cells

Likewise, the identification of Naive CD4+ T cells becomes straightforward by combining CD4 and CCR7:

p4 <- plot_density(pbmc, c("CD4", "CCR7"), joint = TRUE)
p4 + plot_layout(ncol = 1)

Notice that these cells are mainly constrained to cluster 0

p4[[3]] / DimPlot(pbmc, label = TRUE, repel = TRUE)

Conclusions

In summary,Nebulosacan be useful to recover the signal from dropped-out genes and improve their visualization in a two-dimensional space. We recommend using Nebulosa particularly for dropped-out genes. For fairly well-expressed genes, the direct visualization of the gene expression may be preferable. We encourage users to use Nebulosa along with the core visualization methods from the Seurat and Bioconductor environments as well as other visualization methods to draw more informed conclusions about their data.