如何在ggplot2中基于viroid和expression分组进行统计分析？

我正在对一份数据集进行统计分析，该数据集的viroid列包含PSTVd与TPMVd两个取值，expression列包含Up-regulated与Down-regulated两个取值。我已编写如下ggplot2代码绘制箱线图：

ggplot(Both_DEGs, aes(x=Viroid, y=Log2FC, fill=Expression)) +
  geom_boxplot()+
  theme(axis.text.x=NULL)+
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
        panel.background = element_blank(), axis.line = element_line(colour = "black"))+
  stat_compare_means(label = "p.format", 
                     label.x = 1.32, label.y = 18, method = "t.test", paired = F)+
  ggtitle("Expression ranges")+
  theme(plot.title = element_text(hjust = 0.5, size = 14, face = "bold"))

当前生成的箱线图已按Viroid和Expression分为四组，我希望完成两组对比统计分析：PSTVd - Up-regulated vs TPMVd - Up-regulated、PSTVd - Down-regulated vs TPMVd - Down-regulated，请问该如何实现？

你可以通过修改stat_compare_means的参数，指定按Expression分组执行统计检验，同时调整标签位置与箱线组对齐，具体有两种实现方案：

方案1：同一图内分组标注检验结果 #

修改原代码，添加group = Expression让检验按上调/下调组分别执行，同时用position_dodge让标签与对应箱线对齐：

library(ggplot2)
library(ggpubr)

ggplot(Both_DEGs, aes(x = Viroid, y = Log2FC, fill = Expression)) +
  geom_boxplot(position = position_dodge(width = 0.75)) + # 显式设置箱线的分组间距
  theme(axis.text.x = NULL) +
  theme(panel.grid.major = element_blank(), 
        panel.grid.minor = element_blank(),
        panel.background = element_blank(), 
        axis.line = element_line(colour = "black")) +
  stat_compare_means(label = "p.format", 
                     group = Expression, # 按Expression分组执行t检验
                     position = position_dodge(width = 0.75), # 标签与箱线组对齐
                     label.x = c(1.2, 1.2), # 两组标签的x坐标
                     label.y = c(18, 16), # 两组标签的y坐标，避免重叠
                     method = "t.test", 
                     paired = FALSE) +
  ggtitle("Expression ranges") +
  theme(plot.title = element_text(hjust = 0.5, size = 14, face = "bold"))

方案2：分面图分离两组对比 #

如果担心标签重叠，也可以用facet_wrap将上调和下调组拆分为两个子图，每个子图内直接对比两种Viroid的差异：

library(ggplot2)
library(ggpubr)

ggplot(Both_DEGs, aes(x = Viroid, y = Log2FC, fill = Viroid)) +
  geom_boxplot() +
  facet_wrap(~Expression) + # 按Expression分面
  theme(panel.grid.major = element_blank(), 
        panel.grid.minor = element_blank(),
        panel.background = element_blank(), 
        axis.line = element_line(colour = "black")) +
  stat_compare_means(label = "p.format", 
                     method = "t.test", 
                     paired = FALSE) +
  ggtitle("Expression ranges") +
  theme(plot.title = element_text(hjust = 0.5, size = 14, face = "bold"))

两种方案都能实现你需要的分组统计对比，可根据可视化需求选择。

内容的提问来源于stack exchange，提问作者OctavioZM