Chapter 5 Differential gene expression analysis

5.1 Differential gene expression analysis

Let’s now focus on identitying markers that can distinguish the different cell types in the dataset. Let’s use the function FindAllMarkers.

# must switch back to SCT to identify variable markers
Seurat::DefaultAssay(ST011) = "SCT"

# find markers
all_markers = Seurat::FindAllMarkers(
    ST011,
    only.pos = TRUE,
    min.pct = 0.25,
    logfc.threshold = 0.25
)

Let’s check the top markers for each identified.

all_markers %>% group_by(cluster) %>% dplyr::filter(avg_log2FC > 1)

## # A tibble: 961 × 7
## # Groups:   cluster [12]
##        p_val avg_log2FC pct.1 pct.2 p_val_adj cluster gene      
##        <dbl>      <dbl> <dbl> <dbl>     <dbl> <fct>   <chr>     
##  1 0               1.62 0.75  0.409 0         0       MT1F      
##  2 0               1.65 0.878 0.624 0         0       MT1G      
##  3 0               1.04 0.97  0.913 0         0       MT1E      
##  4 3.14e-293       1.58 0.375 0.145 7.04e-289 0       WIF1      
##  5 1.99e-291       1.00 0.728 0.506 4.46e-287 0       ETNPPL    
##  6 1.40e-286       1.12 0.592 0.325 3.14e-282 0       SLC14A1   
##  7 8.73e-276       1.14 0.57  0.311 1.96e-271 0       GJB6      
##  8 2.60e-234       1.49 0.334 0.135 5.82e-230 0       AL121820.2
##  9 2.02e-190       1.00 0.509 0.298 4.52e-186 0       TNS3      
## 10 2.57e-115       1.08 0.276 0.138 5.76e-111 0       TBC1D16   
## # ℹ 951 more rows

Let’s visualize the top 6 in a UMAP plot.

Seurat::FeaturePlot(ST011, features = c("MT1F", "MT1G", "MT1E", "WIF1", "ETNPPL", "SLC14A1"))

Figure 5.1: Top 6 markers

Let’s visualize the top 10 markers that differentiate between clusters with a heatmap.

# get the genes
top10_genes = all_markers %>%
    group_by(cluster) %>%
    dplyr::filter(avg_log2FC > 1) %>%
    slice_head(n = 10) %>%
    ungroup()

# plot
p1 = Seurat::DoHeatmap(ST011, features = top10_genes$gene) + NoLegend()
p1

Figure 5.2: Top markers heatmap

We can also check the distribution of the top 10 markers across the different samples and groups.

Seurat::DotPlot(
    ST011,
    features = unique(top10_genes$gene),
    split.by = "orig.ident",
    cols = c("indianred", "brown2", "firebrick", "#BCD2EE", "lightskyblue3", "dodgerblue")
) + RotatedAxis()

Figure 5.3: Dotplot top markers per sample

Seurat::DotPlot(
    ST011,
    features = unique(top10_genes$gene),
    split.by = "seurat_clusters",
    cols = c(brewer.pal(8, "Set2"), brewer.pal(8, "Dark2"))
) + RotatedAxis()

Figure 5.4: Dotplot top markers per cluster