如何在单柱状图中组合多Y变量并按处理组执行非配对t检验？

需求明确 #

要把CAT、SOD、GSH、MDA四个指标整合为2×2布局的柱状图：

• X轴为Treatments分组，每个处理组对应Pre-challenge和Post-challenge两个时期的柱形
• 针对每个指标，按处理组分别对两个时期做非配对t检验，并用显著性标记（字母或星号）标注差异

数据集 #

CAT   SOD   GSH   MDA   Treatments  Period
2.06  5.18  8.12  2.5   CFP0        Pre-challenge
2.02  5.14  8.18  2.7   CFP0        Pre-challenge
2.68  5.11  8.01  3.12  CFP1        Pre-challenge
2.66  6.15  8.21  3.53  CFP1        Pre-challenge
2.7   5.84  8     4.3   CFP2        Pre-challenge
2.76  5.86  8.2   4.27  CFP2        Pre-challenge
2.92  6.35  7.14  4.79  CFP3        Pre-challenge
2.72  6.35  7.1   4.83  CFP3        Pre-challenge
3.01  7.19  6.36  5     CFP4        Pre-challenge
2.81  7.17  6.42  5.02  CFP4        Pre-challenge
3     8.29  5.13  6.5   CFP5        Pre-challenge
3.2   9.87  5.17  6.58  CFP5        Pre-challenge
5.58  7.12  8.52  7.6   CFP0        Post-challenge
5.4   7.14  8.58  9.2   CFP0        Post-challenge
3.93  4.2   7.4   4.04  CFP1        Post-challenge
3.77  4.5   7.8   4.03  CFP1        Post-challenge
3.81  4.6   7.1   4.2   CFP2        Post-challenge
3.95  4.26  7.86  4.7   CFP2        Post-challenge
3.81  4.68  7.2   4.6   CFP3        Post-challenge
4.2   4.58  7.41  4.7   CFP3        Post-challenge
4.6   6.18  8.3   7.6   CFP4        Post-challenge
4.9   6.09  8.4   7.9   CFP4        Post-challenge
4.9   6.2   8.2   8.96  CFP5        Post-challenge
4.8   6.16  7.9   9.02  CFP5        Post-challenge

现有代码（单变量绘图示例） #

library(readxl)
library(ggpubr)
immun <- read_excel("P_II.xlsx", sheet = "imm_data")
summary(immun)

immun$Period <- as.factor(immun$Period)
immun$Treatments <- as.factor(immun$Treatments)

# 单变量柱状图
imm <- ggbarplot(immun, x = "Treatments", y = c("Lysozyme"), combine = TRUE, 
                 color = "Period", width = 0.5, add = "mean_se", 
                 error.plot = "errorbar", palette = c("#585858","#c6c6c6"), 
                 fill = "Period", position = position_dodge(0.6))
ggpar(imm, ylim = c(120, 190), ylab = "Lysozyme (pg/mL)", title = "Lysozyme")

完整解决方案 #

1. 批量执行非配对t检验 #

用dplyr按处理组分批对每个指标做t检验，先安装加载dplyr：

install.packages("dplyr")
library(dplyr)

# 定义检验函数：输入数据和指标名，返回各处理组的p值
run_tests <- function(data, var_name) {
  data %>%
    group_by(Treatments) %>%
    summarise(
      p_value = t.test(!!sym(var_name) ~ Period, paired = FALSE)$p.value,
      .groups = "drop"
    ) %>%
    mutate(variable = var_name)
}

# 批量处理四个指标
test_results <- bind_rows(
  run_tests(immun, "CAT"),
  run_tests(immun, "SOD"),
  run_tests(immun, "GSH"),
  run_tests(immun, "MDA")
)

# 查看检验结果
print(test_results)

2. 生成显著性标记 #

可以选择星号或字母标注，这里两种方式都提供：

# 安装multcompView包（用于字母标注）
install.packages("multcompView")
library(multcompView)

# 给检验结果添加显著性标记
test_results <- test_results %>%
  mutate(
    # 星号标记
    sig_stars = case_when(
      p_value < 0.001 ~ "***",
      p_value < 0.01 ~ "**",
      p_value < 0.05 ~ "*",
      TRUE ~ "ns"
    ),
    # 字母标记（两个组比较，显著则分a/b，不显著统一a）
    sig_letters = ifelse(p_value < 0.05, "a,b", "a")
  )

3. 绘制2×2布局的多变量柱状图 #

先把数据转成长格式，再用ggpubr批量绘图并添加标注：

install.packages("tidyr")
library(tidyr)

# 转长格式，方便批量绘图
immun_long <- immun %>%
  pivot_longer(cols = c(CAT, SOD, GSH, MDA), names_to = "Variable", values_to = "Value")

# 绘制2×2组合柱状图
combined_plot <- ggbarplot(immun_long, x = "Treatments", y = "Value", 
                           facet.by = "Variable", ncol = 2, # 2×2布局
                           color = "Period", width = 0.5, add = "mean_se", 
                           error.plot = "errorbar", palette = c("#585858","#c6c6c6"), 
                           fill = "Period", position = position_dodge(0.6))

# 添加显著性星号标注（每个处理组的两个柱形上方）
combined_plot <- combined_plot +
  geom_text(data = test_results, 
            aes(x = Treatments, y = max(immun_long$Value)*1.1, label = sig_stars),
            position = position_dodge(0.6), size = 4)

# 调整图表样式
ggpar(combined_plot, 
      ylab = "Indicator Value", 
      title = "Biochemical Indicators: Pre vs Post Challenge",
      legend.title = "Period")

关键提示 #

• 非配对t检验必须设置paired=FALSE，确保和实验设计匹配
• 长格式数据是批量绘制多变量图的核心，pivot_longer能快速完成宽转长
• 若想用字母标注每个时期的差异，可以把sig_letters拆分成两个值，分别对应Pre和Post时期的柱形上方

内容的提问来源于stack exchange，提问作者Muhammad Bello