2
$\begingroup$

I have single cell RNA-seq on 8 time points. I have clustered each time point to 2-3 clusters by seurat. I also have a list of differentially expressed genes between all my 1562 cells in 8 time points. I know how to plot this list of genes in heat map on each cluster of cells or a big heat map of my 1562 cells, but I have also sub-clusters for each cluster (time point).

Could you please help me to figure out how can I have a heat map with sub-clusters for each time point on which I am showing a list of differentially expressed genes?

Something like the picture below where each day has been clustered to 2-3 sub-population then they plotted common differentially expressed genes between day points

Edit

Lets say 500 genes commonly are being differentially expressed in 8 time point. time point 1 has 3 sub-clusters of cells and time points 2 has 2 sub-clusters of cells. time point 3 and 4 both have 2 sub-clusters of cells. I want to heat map these 500 genes in cells in time point 1 (3 sub-clusters), time point 2 (2 sub-clusters), time point 23 (2 sub-clusters) and time point 4 (2 sub-clusters) in one unified heat map on which cells in each subclusteres (a,b and c) are in column and genes in rows but I don't have any idea how to do that. column annotation function in ComplexHeatmap likely works in this regard But I don't know how to define cells inside each sub cluster in different colours

I can heat map of 500 genes only on time points easily by seurat but without underlying sub clusters

> DoHeatmap(object = SubsetData(object = object, max.cells.per.ident =
> 100), genes.use = 50 genes, slim.col.label = TRUE, group.label.rot =
> TRUE)

enter image description here

Figure 2 in this article.

$\endgroup$
  • 4
    $\begingroup$ Some sample expression data to plot in said heatmap will be of use to encourage answers to your question. $\endgroup$ – Matt Bashton Aug 7 '18 at 10:47
  • $\begingroup$ What have you tried to do? Do you have each heatmap for each day and with the same order of genes? $\endgroup$ – llrs Aug 7 '18 at 11:16
  • $\begingroup$ I have independently clustered cells in each time point by seurat. I also obtained differentially expressed genes in each time point separately so I have common genes differentially expressed genes between each pairs of neighbour time points. $\endgroup$ – Angel Aug 7 '18 at 11:32
2
$\begingroup$

In case it is still helpful as the post is rather old, the code below would generate a heatmap with annotations thanks to the ComplexHeatmap package.

But before that I would like to stress out the importance of example data, I bet a lot of experts could not help you just because of they did not have single cell data at their disposal.

I am using some scRNA-seq data stored within a Seurat object. The expression data come from the @scale.data slot and the annotations are from the @meta.data slot. I am using a few general cell type markers as genes and 3 columns from the metadata slot (cell type, cell subtype, sample region). When using the ComplexHeatmap package, the trick is to have the same order in the annotation and actual data tables and this is how you should control what (in your case single cells) is plotted in what order.

library(Seurat)
library(dplyr)
library(ComplexHeatmap)
library(circlize) # for the colorRamp2() function

seurat_object <- readRDS(...)

# Annotations

meta_data <- seurat_object@meta.data

# order of annotations/colors are defined here
ordered_meta_data <- meta_data[order(meta_data$cell_type), ]

# OPTIONAL: YOU CAN PICK COLORS FOR EACH LEVEL OF ANNOTATION
# HERE I PROVIDE COLORS FOR ONLY TWO OF THE THREE LEVELS, COLORS FOR THE
# REMAINING LEVEL IS TAKEN CARE OF BY THE PACKAGE
annotation_colors <- list("cell_type"= c("Epithelial" = "red",
                                         "Immune" = "orange",
                                         "Other" = "gray",
                                         "Stroma" = "green"),
                          "sample_region" = c("B" = "blue",
                                              "C" = "yellow",
                                              "N" = "maroon",
                                              "T" = "darkblue"))
ha = HeatmapAnnotation(df = ordered_meta_data,
                       show_annotation_name = TRUE,
                       col = annotation_colors)

# Expression data

genes_to_use <- c("EPCAM",
                  "CD3D",
                  "NCR1",
                  "CD79A",
                  "LYZ",
                  "COL1A1",
                  "CLDN5",
                  "S100B",
                  "KIT")

seurat_object <- ScaleData(seurat_object,
                           genes.use = genes_to_use)

my_data <- seurat_object@scale.data

# COLUMN ORDER OF THE EXPRESSION DATA SHOULD MATCH THE ROW ORDER OF THE
# ANNOTATION TABLE
my_data <- my_data[, rownames(ordered_meta_data)]

# Heatmap

col_fun = colorRamp2(c(-2, 0, 2), c("blue", "white", "red"))
png(filename = "example_heatmap.png",
    width = 1000,
    height = 1000)
Heatmap(
  my_data,
  col = col_fun,
  cluster_rows = FALSE,
  cluster_columns = FALSE,
  column_order = NULL,
  show_row_dend = FALSE,
  show_column_dend = FALSE,
  show_row_names = TRUE,
  show_column_names = FALSE,
  use_raster = TRUE,
  raster_device = c("png"),
  bottom_annotation = NULL,
  top_annotation = ha
)
dev.off()

enter image description here

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.