CELL_discret.jpg

前言

多组学文章经常出现非连续变量的热图或者叫格子图。举几个例子：

Snipaste_2021-11-25_22-38-06

Snipaste_2021-11-25_22-39-29

以上两个图都来自2021.09的一篇Cell，标题是Proteogenomic characterization of pancreatic ductal adenocarcinoma。今天就不细讲这两幅图了。这种图给我们展示离散/分类变量的差异提供了一个思路。今天就简单介绍几种常用的画这种图的方法。

22

常用方法

构建一个分类变量组成的示例数据。

library(ggplot2)
library(tidyverse)
library(reshape2)
library(RColorBrewer)
clinical.df=data.frame(
  patient=paste("P",seq(1:15),sep = ""),
  age=sample(c("young","old"),15,replace = T),
  gender=sample(c("male","female"),15,replace = T),
  symptom=sample(c("mild","moderate","severe"),15,replace = T),
  RNAseq=sample(c("yes","no"),15,replace = T),
  WES=sample(c("yes","no"),15,replace = T)
)
head(clinical.df)

> head(clinical.df)
  patient   age gender  symptom RNAseq WES
1      P1   old female moderate    yes  no
2      P2   old   male moderate    yes  no
3      P3   old   male moderate    yes yes
4      P4 young female   severe    yes yes
5      P5   old female moderate     no  no
6      P6 young   male moderate     no  no

# 长宽转换 已备作图
clinical.df2=melt(clinical.df,id="patient")
head(clinical.df2)

> head(clinical.df2)
  patient variable value
1      P1      age   old
2      P2      age   old
3      P3      age   old
4      P4      age young
5      P5      age   old
6      P6      age young

geom_tile

Color<-brewer.pal(9, "Set3") # 设置颜色
# 设置因子顺序
clinical.df2$patient=factor(clinical.df2$patient,levels = paste("P",seq(1:15),sep = ""))
clinical.df2$variable=factor(clinical.df2$variable,levels = c("WES","RNAseq","symptom","gender","age"))
ggplot(clinical.df2, aes(x = patient, y = variable, fill = value)) +
  geom_tile(color = "white", size = 0.25) +
  scale_fill_manual(name = "Category",
                    #labels = names(sort_table),
                    values = Color)+
  theme(#panel.border = element_rect(fill=NA,size = 2),
    panel.background = element_blank(),
    plot.title = element_text(size = rel(1.2)),
    axis.title = element_blank(),
    axis.ticks = element_blank(),
    legend.title = element_blank(),
    legend.position = "right")

image-20211125225237494

ggwaffle

devtools::install_github("liamgilbey/ggwaffle") # 下载包
library(ggwaffle)
ggplot(clinical.df2, aes(patient, variable, fill = value)) + 
  geom_waffle()+
  scale_fill_manual(name = "Category",
                    #labels = names(sort_table),
                    values = Color)+
  theme(#panel.border = element_rect(fill=NA,size = 2),
    panel.background = element_blank(),
    plot.title = element_text(size = rel(1.2)),
    axis.title = element_blank(),
    axis.ticks = element_blank(),
    legend.title = element_blank(),
    legend.position = "right")

和geom_tile异曲同工。

image-20211125225912215

ComplexHeatmap

ComplexHeatmap应该是最能还原本文前言图的包，不过我这里暂时还没时间搞定，后续发复现版本的代码。

row.names(clinical.df) <- clinical.df[,1]
clinical.df <- clinical.df[,-1]
clinical.df3 <- data.frame(t(clinical.df)) 
# 上面的代码为了将数据转为热图矩阵
library(ComplexHeatmap)
Heatmap(clinical.df3)

image-20211125230958807

未经雕饰的图确实不是很美观。

总结

以上就是我所知的几种常用的画离散变量的热图的方法。如果大家有更巧妙的想法，欢迎在后台留言互相学习交流。

参考

R绘图(2): 离散/分类变量如何画热图/方块图 - 简书 (jianshu.com)

往期

1. 跟着Nature学作图 | 配对哑铃图+分组拟合曲线+分类变量热图
2. (免费教程+代码领取)|跟着Cell学作图系列合集
3. 跟着Nat Commun学作图 | 1.批量箱线图+散点+差异分析
4. 跟着Nat Commun学作图 | 2.时间线图
5. 跟着Nat Commun学作图 | 3.物种丰度堆积柱状图
6. 跟着Nat Commun学作图 | 4.配对箱线图+差异分析