复现PNAS图表：Complexheatmap/ggplot2绘制离散型热图及修饰

KS科研分享与服务-TS的美梦

发布于 2025-12-20 17:07:53

100

使用Complexheatmap/ggplot2绘制离散型热图，复现PNAS图表！

文章原图：

（reference：Convergent molecular, cellular, and cortical neuroimaging signatures of major depressive disorder）

完整的复现效果：

读入数据

setwd("/Users/ks_ts/Documents/公众号文章/2022-11-22Complexheatmap热图聚类注释/")
Enrich_df<- read.csv("./Enrich_df.csv", header = T, row.names = 1)
library(dendextend) 
library(ComplexHeatmap)
library(tibble)
library(ggplot2)
head(Enrich_df)

行聚类信息

Enrich_df1 <- as.data.frame(t(Enrich_df))#转置
group_info <- as.data.frame(rownames(Enrich_df1))#做一个分组信息数据
dend1 = cluster_between_groups(Enrich_df, group_info$`rownames(Enrich_df1)`)#聚类
dend1 = color_branches(dend1, k = 4, col = c("#F5A089",'#CBA53C','#88CFC8','#66A5CA'))#设置聚类数，颜色  
dend1 = dend1 %>% set("branches_lwd", 2) #聚类树的宽度
dend1

查看聚类情况

HM <- Heatmap(as.matrix(Enrich_df1),cluster_rows = dend1, cluster_columns = F)#做一个热图看看效果
HM

画热图


#颜色设置,需要是原始行名顺序
group_info$cluser = cutree(dend1, k = 4)
group_info$color = ifelse(group_info$cluser=='1', "#F5A089",
                          ifelse(group_info$cluser=='2','#CBA53C',
                                 ifelse(group_info$cluser=='3','#88CFC8','#66A5CA')))


col_cluster = group_info$color 
names(col_cluster) <- group_info$`rownames(Enrich_df1)`

#-----作图
Heatmap(as.matrix(Enrich_df1),
        cluster_rows = dend1,#行聚类用前面设置好的、这样颜色也对应设置
        cluster_columns = F,
        show_column_names = T,
        show_row_names = T,
        row_names_side =  'left',
        column_title = NULL,
        heatmap_legend_param = list(
          title=' ',
          labels=c('Present','Absent'),
          direction = "horizontal",
          labels_gp = gpar(fontsize = 10),
          border='black'), #legend设置

        col = c('#F2F2F0','#5A8FCA'),
        rect_gp = gpar(col = "grey", lwd = 1),

        row_names_gp = gpar(col=col_cluster, fontsize=10),#设置文字颜色

        column_names_gp = gpar(fontsize = 10))

不过学习重在举一反三，我突然想到，有一点内容是可以用到和拓展的，那就是对于行列的注释，之前我们一直使用图形颜色注释，看起来费劲，那么是否可以直接将文字注释为不用的颜色呢？显然可以：

row_info = rowAnnotation(foo = anno_text(rownames(Enrich_df1), 
                                   location = 0, 
                                   just = "left",
                                   gp = gpar(col=col_cluster, fontsize=10)))

Heatmap(as.matrix(Enrich_df1),
        cluster_rows = T,
        cluster_columns = F,
        show_column_names = T,
        show_row_names = F,
        column_title = NULL,
        heatmap_legend_param = list(
          title=' ',
          labels=c('Present','Absent'),
          labels_gp = gpar(fontsize = 10),
          border='black'),
        col = c('#F2F2F0','#5A8FCA'),
        rect_gp = gpar(col = "grey", lwd = 1),
        row_names_gp = gpar(col=col_cluster, fontsize=10),
        column_names_gp = gpar(fontsize = 10))+
  row_info

这样就完成注释了，不同的颜色代表不同组！

ggplot2绘制

到这里，估计还有小伙伴想问ggplot2做怎么弄，就简单演示一下吧：ggplot2同样能够完成热图绘制，只不过没有那么灵活，需要将宽数据转化为长数据：

library(tidyr)
library(dplyr)

# 转换为长格式
df_long <- Enrich_df1 %>%
  rownames_to_column("pathway") %>%
  pivot_longer(cols = -pathway, names_to = "gene", values_to = "value")

#将value转化为因子，固定x轴基因顺序
df_long$value <- as.factor(df_long$value)
df_long$gene <- factor(df_long$gene,levels=colnames(Enrich_df1))


#plot
ggplot(data=df_long,aes(x=gene,y=pathway))+  
  geom_tile(aes(fill=value),color="grey")+  
  theme_minimal()+  
  theme(panel.border = element_rect(fill=NA,color="black", size=1, linetype="solid"),        
        panel.grid = element_blank(),        
        axis.ticks.y = element_blank(),        
        axis.title = element_blank(),        
        axis.text.x = element_text(angle=90,hjust=1, vjust=0.5,colour = 'black', size = 10),       
        axis.text.y = element_text(colour = 'black', size = 10),        
        plot.margin=unit(c(0.4,0.4,0.4,0.4),units=,"cm"))+  
  scale_fill_manual(values = c('#F2F2F0','#5A8FCA'))

绘制聚类：

library(ggtree)
##行聚类
phr <- hclust(dist(Enrich_df1),method = "ward.D2") 

#plot聚类树
p_cluster <- ggtree(phr, branch.length ="none")
p_cluster

组合：

#将value转化为因子，固定x轴基因顺序
df_long$value <- as.factor(df_long$value)
df_long$gene <- factor(df_long$gene,levels=colnames(Enrich_df1))

#plot
p = ggplot(data=df_long,aes(x=gene,y=pathway))+  
  geom_tile(aes(fill=value),color="grey")+  
  theme_minimal()+  
  theme(panel.border = element_rect(fill=NA,color="black", size=1, linetype="solid"),        
        panel.grid = element_blank(),        
        axis.ticks.y = element_blank(),        
        axis.title = element_blank(),        
        axis.text.x = element_text(angle=90,hjust=1, vjust=0.5,colour = 'black', size = 10),       
        axis.text.y = element_text(colour = 'black', size = 10),        
        plot.margin=unit(c(0.4,0.4,0.4,0.4),units=,"cm"))+  
  scale_fill_manual(
    values = c("0" = '#F2F2F0', "1" = '#5A8FCA'),
    labels = c("0" = "Absent", "1" = "Present"),  # 修改legend标签
    name = ""                        # 修改legend标题
  )

library(aplot)
p %>% insert_left(p_cluster, width =0.2)