Chapter 7 ChipHeatmap based on grid

Here I developed ChipHeatmap and multiHeatmap functions to visualize the heatmap and profile plot based on grid and gtable R package. ChipHeatmap mainly draw single plot and multiHeatmap can combine multiple plot together. Now we show some examples here.

---

7.1 Load data

You can load the nomalized matrix data which from normalizeToMatrix/getTagMatrix/parseDeeptools output, here we use parseDeeptools output data which is produced by deeptools software:

library(ggChIPvis)
library(grid)
library(gtable)

# parse deeptools output
deep_mat <- parseDeeptools(deeptools_output = "refPoint-data.gz")

row_split <- rep(c("group 1","group 2","group 3"),c(500,1500,3000))

7.2 Single heatmap visualization

Default output:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]])
popViewport()

Rotate the xaxis:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],ht.xaxis.rot = 45)
popViewport()

Control the sample label style:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],
            sample.label = "jinitaimei",
            label.rect.gp = gpar(fill = "pink"),
            label.gp = gpar(fontface = "bold.italic"))
popViewport()

Control the panel space:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],
            panel.col.space = 0,panel.row.space = 0)
popViewport()

Remove extreme values:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],quantile.threshold = 0.99)
popViewport()

Control the heatmap colors:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],
            ,quantile.threshold = 0.99,
            ht.col = c("grey50","green","red"))
popViewport()

Change the shape of the left anno bar:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],draw.anno.left.shape = "rect")
popViewport()

Change the legend position:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],
            quantile.threshold = 0.99,
            legend.pos = "bottom")
popViewport()

Add profile plot:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],draw.profile = T)
popViewport()

Split the rows:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],draw.profile = T,
            row.split = row_split)
popViewport()

Change the left annobar shape:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],draw.profile = T,
            row.split = row_split,
            draw.anno.left.shape = "rect")
popViewport()

Keep the cluster height size to be same:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],draw.profile = T,
            row.split = row_split,
            keep.cluster.panel.same = T)
popViewport()

Control the size of each panel:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],draw.profile = T,
            HeatmapLayout.params = list(heatmap.profile.h = 0.4,
                                        heatmap.size = c(0.5,0.5)))
popViewport()

Change the left annobar direction:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],
            draw.anno.left.diretion = "up")
popViewport()

Change the profile line colors, cluster panel gap and left annobar colors:

grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = deep_mat[[1]],
            row.split = row_split,
            keep.cluster.panel.same = T,
            draw.profile = T,
            draw.anno.left.shape = "rect",
            anno.left.col = RColorBrewer::brewer.pal(3,"Set3"),
            profile.line.col = RColorBrewer::brewer.pal(3,"Set3"),
            cluster.gap = 0.05)
popViewport()

7.3 Plot getTagMatrix output

library(ChIPseeker)
library(TxDb.Hsapiens.UCSC.hg19.knownGene)
txdb <- TxDb.Hsapiens.UCSC.hg19.knownGene

files <- getSampleFiles()

seek_list <- lapply(4:5, function(x){
  peak <- readPeakFile(files[[x]])
  
  ## The results of binning method and normal method are nearly the same. 
  tagMatrix_binning <- getTagMatrix(peak = peak, TxDb = txdb, 
                                    upstream = 3000, downstream = 3000, 
                                    type = "body", by = "gene", 
                                    nbin = 800)
})


grid.newpage()
pushViewport(viewport(x = 0.5,y = 0.5,width = 0.5,height = 0.85))
ChipHeatmap(norm.mat = seek_list[[1]],
            heatmap_rank_method = "sum",
            draw.profile = T)
popViewport()