RStudio中heatmap.2颜色键不显示问题求助

问题描述 #

在RStudio中绘制heatmap.2热力图时，颜色键（color key）突然消失，尝试切换调色板、注释边距相关代码后问题仍未解决。原代码如下：

ghe <- read.csv("greenhouse-emissions.csv")

gh_df <- ghe[ghe$Pollutant == "Greenhouse gases",]
gh_df <- gh_df[gh_df$VAR == "TOTAL", ]

cyv_df <- gh_df[c('Country', 'Year', 'Value')]
cyv_df <- cyv_df[cyv_df$Country != "OECD - Total",]

new_df <- cyv_df %>% 
  pivot_wider(names_from = Year, values_from = Value)

new_df[is.na(new_df)] <- 0

#Creating matrix for heatmap.2
m <- data.matrix(new_df)
m <- m[,-1]

# My original color palette. However, I later decided to not use the palette because I thought the palette was the cause behind the color key disappearance.
my_palette <- colorRampPalette(c("mintcream",
                                 "cadetblue",
                                 "darkolivegreen1",
                                 "yellow", 
                                 "tan1",
                                 #"darkgoldenrod",
                                 "red",
                                 "red2",
                                 "red4",
                                 "black"))(99)

col_breaks = c(seq(-1, 1, length=1),
               seq(1.001, 3428.207,length=5),
               seq(3428.208, 61257.344, length=15),
               seq(61257.345, 592587.216,length=10), #upper smaller countries
               seq(592587.217, 1.27e+06, length=19), #Germany
               seq(1270000.001, 1.33e+06, length=19),# Japan
               seq(1330000.001, 2.5e+06,length=11), # Russia
               seq(2500000.001, 6.25e+06, length=10),
               seq(6250000.001, 17124526.000, length=10))

Colors=rev(brewer.pal(11,"Spectral"))
Colors=colorRampPalette(Colors)(99)

dev.off()
heatmap.2(m,
          col=Colors, 
          breaks=col_breaks,
          trace="none", 
          Colv=NA, 
          # Rowv = NA,
          dendrogram = 'none',
          margins=c(5,5),
          density.info='histogram', 
          denscol="black",
          key=TRUE,
          symkey=FALSE,
          keysize=1, 
          key.par=list(mar=c(3.5,0,3,0)),
          lmat=rbind(c(5, 4, 2), c(6, 1, 3)), 
          lwid=c(1, 10, 1),
          notecol = 'black')

修复方案 #

1. 调整布局参数 #

自定义的lmat和lwid布局可能挤压了颜色键的显示空间：

• 先注释掉lmat和lwid参数，使用默认布局测试颜色键是否恢复；
• 若需保留自定义布局，修改lwid的宽度比例，比如lwid=c(2,10,1)，给颜色键所在列留出足够宽度。

2. 匹配颜色与断点数量 #

heatmap.2要求颜色数量等于length(col_breaks)-1。当前代码中Colors有99种颜色，但col_breaks长度仅为90，导致参数不匹配引发异常。修改col_breaks的length.out参数，确保length(col_breaks)为100（对应99种颜色）：

col_breaks <- c(seq(-1, 1, length.out=2),
               seq(1.001, 3428.207, length.out=6),
               seq(3428.208, 61257.344, length.out=16),
               seq(61257.345, 592587.216, length.out=11),
               seq(592587.217, 1.27e+06, length.out=20),
               seq(1270000.001, 1.33e+06, length.out=20),
               seq(1330000.001, 2.5e+06, length.out=12),
               seq(2500000.001, 6.25e+06, length.out=11),
               seq(6250000.001, 17124526.000, length.out=11))

3. 修改颜色键边距 #

key.par=list(mar=c(3.5,0,3,0))中左右边距设为0，可能导致颜色键被截断，调整为：

key.par=list(mar=c(3.5,1,3,1))

修改后的完整测试代码 #

ghe <- read.csv("greenhouse-emissions.csv")

gh_df <- ghe[ghe$Pollutant == "Greenhouse gases",]
gh_df <- gh_df[gh_df$VAR == "TOTAL", ]

cyv_df <- gh_df[c('Country', 'Year', 'Value')]
cyv_df <- cyv_df[cyv_df$Country != "OECD - Total",]

new_df <- cyv_df %>% 
  pivot_wider(names_from = Year, values_from = Value)

new_df[is.na(new_df)] <- 0

#Creating matrix for heatmap.2
m <- data.matrix(new_df)
m <- m[,-1]

Colors=rev(brewer.pal(11,"Spectral"))
Colors=colorRampPalette(Colors)(99)

# 修正后的断点，匹配颜色数量
col_breaks <- c(seq(-1, 1, length.out=2),
               seq(1.001, 3428.207, length.out=6),
               seq(3428.208, 61257.344, length.out=16),
               seq(61257.345, 592587.216, length.out=11),
               seq(592587.217, 1.27e+06, length.out=20),
               seq(1270000.001, 1.33e+06, length.out=20),
               seq(1330000.001, 2.5e+06, length.out=12),
               seq(2500000.001, 6.25e+06, length.out=11),
               seq(6250000.001, 17124526.000, length.out=11))

dev.off()
heatmap.2(m,
          col=Colors, 
          breaks=col_breaks,
          trace="none", 
          Colv=NA, 
          dendrogram = 'none',
          margins=c(5,5),
          density.info='histogram', 
          denscol="black",
          key=TRUE,
          symkey=FALSE,
          keysize=1, 
          key.par=list(mar=c(3.5,1,3,1)),
          # 先注释自定义布局，测试后再按需调整
          # lmat=rbind(c(5, 4, 2), c(6, 1, 3)), 
          # lwid=c(2, 10, 1),
          notecol = 'black')

内容的提问来源于stack exchange，提问作者SH Yu