Heatmap in R; how to change the order of Row names

Question

[hellokwmin]

Hello, Biostars,

I am working on generation of Heatmap.

My R code used for that is:

Heatmap(heat,
        name = "Log2",
        col = color_fun,
        split = df$Family,
        cluster_rows = T,
        show_row_dend = F,
        show_row_names = F,
        row_title_side = "left",
        row_title_gp = gpar(fontsize = 12, fontface = "bold"),
        row_names_side = "left",
        row_title_rot = 0,
        width = unit(4, "cm"))

And then, I have this figure:

Since I run cluster_row = T function, the order of row is not from 1 to 9. How could I make it from 1 to 9 keeping clustering?

Thank you.

[Eshwari]

Hi, One reason this is happening is because of the dendogram. Disable the dendogram for the columns and then check. In most usual cases it sorts the rows order wise

[foejvs546]

Alternatively you could use ggplots geom_tile to make the heatmap. Then you could set facet_wrap(~Family), and change the order by setting the factor

df$Family = factor(df$Family, levels=c( ....   ))

Answer 1

[bk11]

You can add cluster_row_slices = F to your code.

head(heat)
C1         C2
R1 -0.56047565 -0.6947070
R2 -0.23017749 -0.2079173
R3  1.55870831 -1.2653964
R4  0.07050839  2.1689560
R5  0.12928774  1.2079620
R6  1.71506499 -1.1231086

Heatmap(heat, name = "Log2", col = color_fun, split = df$Family,  
         cluster_rows = T, show_row_dend = F, show_row_names = F,   
         row_title_side = "left", row_title_gp = gpar(fontsize = 12, fontface = "bold"), 
         row_names_side = "left", row_title_rot = 0, width = unit(4, "cm"), cluster_row_slices = F)

OR you can do km=9 (I m seeing 9 row clusters in your figure) and modify your code like this:-

Heatmap(heat, name = "Log2", col = color_fun,  cluster_rows = T, 
                show_row_dend = F, show_row_names = F, row_title_side = "left", 
                row_title_gp = gpar(fontsize = 12, fontface = "bold"), row_names_side = "left", 
                row_title_rot = 0, width = unit(4, "cm"), km=9, cluster_row_slices = F)