Filter top differentially experssed for pheatmap from DESeq2

Question

[synat.keam]

Dear All experts,

Hope you are all well. I am doing hierachical clustering to see whether sample in each group is clustered together. I used DESeq2 model and my code chunk below to do the clustering. I am wondering if I wanted to filter say top 100 differentially expressed genes, where should I add to my existing code chunk. Looking forward to hearing from you soon. This is just example dataset.

Kind Regards,

Synat

## Heatmap 

sampleDists <- dist(t(assay(vsd)))
library("RColorBrewer")
sampleDistMatrix <- as.matrix(sampleDists)
rownames(sampleDistMatrix) <- paste(vsd$Mouse, sep="-")
colnames(sampleDistMatrix) <- NULL
colors <- colorRampPalette( rev(brewer.pal(9, "Blues")))(255)

## annotation ============================ 

annotation.row <- data.frame(Treatment = factor(vsd@colData$Treatment,
                                                levels = c("Radiotherapy", "Sham-radiotherapy")), 
                             Group = factor(vsd@colData$Group,
                                            levels = c("0Gy:0.1 day","0Gy:1 day","0Gy:4 days",
                                                       "0Gy:8 days", "2Gy×5:0.1 day", "2Gy×5:1 day", 
                                                       "2Gy×5:4 days", "2Gy×5:8 days")))


## mapping =======================
rownames(annotation.row)<- rownames(sampleDistMatrix)

# plot the pheatmap ====================
library(pheatmap)
pheatmap(sampleDistMatrix,
         clustering_distance_rows=sampleDists,
         clustering_distance_cols=sampleDists,
         annotation_row = annotation.row, 
         border_color = TRUE, cluster_cols = F)

[ATpoint]

Scaling RNA-Seq data before clustering?#

You would extract the DEGs from the results DESeq2 object (based on padj or logFC, or both) and then subset the vsd object with it, then scaling and plotting the heatmap.