Complex Heatmap

In this supplemental material, we are creating the same published figure, shown below, using ComplexHeatmap.

The first action we need to perform is a scaling of our counts data. There is no row scaling argument in ComplexHeatmap, so we need to perform the scaling manually prior to creating the heatmap.

# Manually scale and center the data
mat <- heatmap_normCounts
mat_scaled = t(apply(mat, 1, scale))
colnames(mat_scaled) <- colnames(heatmap_normCounts)

Then we need to make sure the metadata contains the same columns as the counts data (in the same order).

# Reorder metadata to match counts
mat_meta <- heatmap_meta[match(colnames(mat_scaled), rownames(heatmap_meta)), , drop = FALSE]

Now create the heatmap using the Heatmap() function. There are so many arguments to customize and an incredible amount of functionality for adding additional plots alongside the heatmap. The ComplexHeatmap book is suggested reading whenever creating a heatmap with this package.

# Check options
?Heatmap

# Generate heatmap
hp <- Heatmap(mat_scaled,
              col = heatmap_colors,
              column_km = 2,
              row_km = 2,
              column_title = gt_render(c("*Prdm16* cKO", "*Prdm16* WT")),
              row_title = c("Up-regulated genes", "Down-regulated genes"),
              show_row_names = FALSE,
              show_column_names = FALSE,
              column_title_gp = gpar(fontsize = 14, fontface = "italic"),
              column_dend_reorder = c(4, 3, 2, 1, 2, 3, 1, 4),
              show_parent_dend_line = FALSE,
              row_gap = unit(0, "mm"),
              column_gap = unit(0, "mm"),
              heatmap_legend_param = list(
                title = "", at = c(-2, -1, 0, 1, 2)),
              top_annotation = HeatmapAnnotation(Genotype = anno_simple(mat_meta$genotype,
                                                             col = c(KO = "#F8766D",
                                                                     WT = "#00BFC4")),
                                                 annotation_name_side = "left"))

heatmap <- draw(hp,
     column_title_gp = gpar(fontsize = 16))

Now to add additional annotations, we can change the image into a ggplot2 object using ggplotify and then use cowplot to add the annotations.

# Change into ggplot object
gg_heatmap <- as.ggplot(hp)


# Add title with cowplot
ggdraw(gg_heatmap) +
  draw_label("Radial glia",
             x = 0.45,
             y = 0.98,
             size = 16,
             hjust = 0,
             vjust = 0,
             fontface = "plain")