如何为ggplot中每个变量设置独立颜色刻度？

你当前的代码生成的热图会让所有变量共享同一色阶，但不同变量（比如连续型的Age、多分类的Tumor_Stage、二分类的Smoking_Status）的数值范围和含义差异很大，需要给每个变量单独配置颜色刻度。可以借助ggnewscale包的new_scale_fill()和new_scale_color()函数实现这个需求，核心思路是分图层绘制每个变量，每个图层单独设置颜色映射。

步骤1：安装并加载必要包 #

首先确保安装了ggnewscale和tidyverse相关包：

install.packages(c("tidyverse", "ggnewscale"))
library(tidyverse)
library(ggnewscale)

步骤2：完整实现代码（手动分层版本） #

这种方式适合变量较少的场景，可以精细控制每个变量的颜色方案：

# 构造原始数据集
Case <- c("Case 1", "Case 2", "Case 3", "Case 4", "Case 5")
Age <- c(53, 46, 72, 68, 45)
Tumor_Stage <- c(1, 2, 3, 1, 2) 
Tumor_Grade <- c(3, 1, 2, 2, 1)
Smoking_Status <- c(0,1 ,1 ,0 ,1)
CD3 <- c(0,1,0,0,1)
df <- tibble(Case, Age, Tumor_Stage, Tumor_Grade, Smoking_Status, CD3)

# 初始化画布，逐个添加变量图层
ggplot() +
  # 1. Age变量（连续型，渐变色）
  geom_tile(data = df %>% mutate(Variables = "Age"),
            aes(x = Case, y = Variables, fill = Age, color = Age)) +
  scale_fill_gradient(name = "Age", low = "#f7fbff", high = "#08306b") +
  scale_color_gradient(name = "Age", low = "#f7fbff", high = "#08306b") +
  
  # 重置填充和颜色刻度，为下一个变量准备
  new_scale_fill() +
  new_scale_color() +
  
  # 2. Tumor_Stage变量（多分类，离散色）
  geom_tile(data = df %>% mutate(Variables = "Tumor_Stage"),
            aes(x = Case, y = Variables, fill = factor(Tumor_Stage), color = factor(Tumor_Stage))) +
  scale_fill_discrete(name = "Tumor Stage") +
  scale_color_discrete(name = "Tumor Stage") +
  
  # 重置刻度
  new_scale_fill() +
  new_scale_color() +
  
  # 3. Tumor_Grade变量（多分类，自定义离散色）
  geom_tile(data = df %>% mutate(Variables = "Tumor_Grade"),
            aes(x = Case, y = Variables, fill = factor(Tumor_Grade), color = factor(Tumor_Grade))) +
  scale_fill_viridis_d(name = "Tumor Grade", option = "plasma") +
  scale_color_viridis_d(name = "Tumor Grade", option = "plasma") +
  
  # 重置刻度
  new_scale_fill() +
  new_scale_color() +
  
  # 4. Smoking_Status变量（二分类，手动配色）
  geom_tile(data = df %>% mutate(Variables = "Smoking_Status"),
            aes(x = Case, y = Variables, fill = factor(Smoking_Status), color = factor(Smoking_Status))) +
  scale_fill_manual(name = "Smoking Status", values = c("0" = "#e6f5c9", "1" = "#386641"), 
                    labels = c("Non-smoker", "Smoker")) +
  scale_color_manual(name = "Smoking Status", values = c("0" = "#e6f5c9", "1" = "#386641"), 
                     labels = c("Non-smoker", "Smoker")) +
  
  # 重置刻度
  new_scale_fill() +
  new_scale_color() +
  
  # 5. CD3变量（二分类，手动配色）
  geom_tile(data = df %>% mutate(Variables = "CD3"),
            aes(x = Case, y = Variables, fill = factor(CD3), color = factor(CD3))) +
  scale_fill_manual(name = "CD3 Expression", values = c("0" = "#fee0d2", "1" = "#de2d26"), 
                    labels = c("Negative", "Positive")) +
  scale_color_manual(name = "CD3 Expression", values = c("0" = "#fee0d2", "1" = "#de2d26"), 
                     labels = c("Negative", "Positive")) +
  
  # 调整主题和轴标签
  labs(x = "Case", y = "Variables") +
  theme_minimal()

步骤3：优化版本（循环批量处理） #

如果变量数量较多，用循环可以减少重复代码，提高可维护性：

# 定义变量配置列表：包含变量名、类型、颜色刻度
var_config <- list(
  list(name = "Age", type = "continuous", 
       fill_scale = scale_fill_gradient(name = "Age", low = "#f7fbff", high = "#08306b"),
       color_scale = scale_color_gradient(name = "Age", low = "#f7fbff", high = "#08306b")),
  list(name = "Tumor_Stage", type = "categorical", 
       fill_scale = scale_fill_discrete(name = "Tumor Stage"),
       color_scale = scale_color_discrete(name = "Tumor Stage")),
  list(name = "Tumor_Grade", type = "categorical", 
       fill_scale = scale_fill_viridis_d(name = "Tumor Grade", option = "plasma"),
       color_scale = scale_color_viridis_d(name = "Tumor Grade", option = "plasma")),
  list(name = "Smoking_Status", type = "binary", 
       fill_scale = scale_fill_manual(name = "Smoking Status", values = c("0"="#e6f5c9", "1"="#386641"), 
                                      labels = c("Non-smoker", "Smoker")),
       color_scale = scale_color_manual(name = "Smoking Status", values = c("0"="#e6f5c9", "1"="#386641"), 
                                       labels = c("Non-smoker", "Smoker"))),
  list(name = "CD3", type = "binary", 
       fill_scale = scale_fill_manual(name = "CD3 Expression", values = c("0"="#fee0d2", "1"="#de2d26"), 
                                      labels = c("Negative", "Positive")),
       color_scale = scale_color_manual(name = "CD3 Expression", values = c("0"="#fee0d2", "1"="#de2d26"), 
                                       labels = c("Negative", "Positive")))
)

# 初始化画布
p <- ggplot()

# 循环添加每个变量的图层和刻度
for (i in seq_along(var_config)) {
  var <- var_config[[i]]
  # 准备当前变量的数据：添加Y轴标签，分类变量转因子
  data_layer <- df %>% 
    mutate(Variables = var$name,
           value = .data[[var$name]]) %>%
    {if (var$type != "continuous") mutate(., value = factor(value)) else .}
  
  # 添加热图块
  p <- p + geom_tile(data = data_layer, aes(x = Case, y = Variables, fill = value, color = value))
  
  # 添加颜色刻度
  p <- p + var$fill_scale + var$color_scale
  
  # 非最后一个变量时，重置刻度系统
  if (i != length(var_config)) {
    p <- p + new_scale_fill() + new_scale_color()
  }
}

# 调整主题和标签
p <- p + labs(x = "Case", y = "Variables") + theme_minimal()

# 输出图表
print(p)

核心逻辑说明 #

• 每个变量单独作为一个geom_tile图层，通过mutate生成统一的Variables列来对应Y轴位置
• 连续变量用scale_fill_gradient设置渐变色，分类/二分类变量用scale_fill_discrete或scale_fill_manual设置离散色
• new_scale_fill()和new_scale_color()会重置当前的颜色映射系统，让后续图层的颜色配置完全独立于之前的变量
• 可以根据变量的实际含义自定义颜色，比如二分类变量用对比色，连续变量用渐变色，提升图表的可读性

内容的提问来源于stack exchange，提问作者Alphonse Charbel