如何解决使用ggplot与ggsignif的geom_signif函数时的x轴分组报错问题

geom_signif默认需要读取全局x轴的分组美学映射，你当前的代码未在ggplot()的全局参数中定义x映射，而是为各个图层单独手动指定x坐标，因此geom_signif无法自动识别分组逻辑，触发报错。

最简单的修复方式是在geom_signif参数中添加manual = TRUE，开启手动指定标记位置的模式，无需修改其他原有代码。

修复后的完整可运行代码如下：

library(ggplot2)
library(ggsignif)
library(ggh4x) # 用于axis_minor坐标轴设置

d <- data.frame(y = rnorm(20, 9, 2),
                group = as.factor(rep(c('Post-FAP', 'Post-DEP'), each = 10)),
                id = rep(1:10, 2))

ggplot(d, aes(y = y)) +
  geom_boxplot(aes(x = rep(c(-2.5, 2.5), each = 10), group = group), fill = '#47E3FF') +
  geom_point(aes(x = rep(c(-1, 1), each = 10)), shape = 21, size = 1.5, col = "black", fill = "grey") +
  geom_line(aes(x = rep(c(-1, 1), each = 10), group = id)) +
  geom_signif(annotation = "p=0.05", y_position = 13, xmin = -2.5, xmax = 2.5, tip_length = .02, manual = TRUE) +
  scale_x_continuous(breaks = c(-2.5, 2.5), labels = c("Post-FAP", "Post-DEP")) +
  scale_y_continuous(minor_breaks = seq(5, 14, by =1),
                     breaks = seq(6, 14, by = 2), limits = c(5, 14),
                     guide = "axis_minor") + 
  theme_bw() +
  theme(legend.position = "none", panel.grid = element_blank())

如果你希望代码更符合ggplot的语法规范，也可以直接将group映射到x轴，再通过position_nudge调整各图层的偏移位置，后续添加更多显著性标记时会更方便：

ggplot(d, aes(x = group, y = y)) +
  geom_boxplot(position = position_nudge(x = c(-1.5,1.5)), fill = '#47E3FF', width = 1) +
  geom_point(position = position_nudge(x = c(-1,1)), shape = 21, size = 1.5, col = "black", fill = "grey") +
  geom_line(aes(group = id), position = position_nudge(x = c(-1,1))) +
  geom_signif(annotation = "p=0.05", y_position = 13, xmin = 1, xmax = 2, tip_length = .02) +
  scale_x_discrete(labels = c("Post-FAP", "Post-DEP")) +
  scale_y_continuous(minor_breaks = seq(5, 14, by =1),
                     breaks = seq(6, 14, by = 2), limits = c(5, 14),
                     guide = "axis_minor") + 
  theme_bw() +
  theme(legend.position = "none", panel.grid = element_blank(),
        axis.title.x = element_blank())

内容的提问来源于stack exchange，提问作者alunats