ggplot2图例显示异常：如何修复非圆形图例？

问题出在你的geom_text继承了全局的color = type映射，导致ggplot把文本元素也纳入了图例，所以图例里不是单纯的圆形点，而是点加文本的组合。

有两种简单的修复方式：

方式一：让geom_text不显示图例 #

给两个geom_text都加上show.legend = FALSE参数，明确告诉ggplot不要把文本元素放进图例：

control <- read.table(text = "genename  fc  p-val   q-val   type
a   0.4 0.0034  0.031   con
b   2   0.0005  0.01    con", header = T,sep="\t")

treatment <- read.table(text = "genename    fc  p-val   q-val   type
a   0.6 0.002   0.01    treat
c   1.5 0.000066    0.004   treat", header =T, sep="\t")


library(ggplot2)
ggplot(control,aes(x = log2(fc),
                   y = -log10(q.val), 
                   color = type))+
  geom_point() + 
  geom_text(data = subset(control, genename == "a"),
            aes(label = genename, vjust = 2),
            show.legend = FALSE) +  # 添加这行
  geom_point(data = treatment) + 
  geom_text(data = subset(treatment, genename == "a"),
            aes(label = genename, vjust = 2),
            show.legend = FALSE) +  # 添加这行
  scale_color_manual(values = c("con" = "blue",
                                "treat" = "red"))

方式二：把颜色映射仅绑定给geom_point #

把color = type从全局的ggplot() aes里移到geom_point的aes中，这样只有点会使用颜色映射，文本不会继承这个映射，也就不会干扰图例：

control <- read.table(text = "genename  fc  p-val   q-val   type
a   0.4 0.0034  0.031   con
b   2   0.0005  0.01    con", header = T,sep="\t")

treatment <- read.table(text = "genename    fc  p-val   q-val   type
a   0.6 0.002   0.01    treat
c   1.5 0.000066    0.004   treat", header =T, sep="\t")


library(ggplot2)
ggplot(control,aes(x = log2(fc),
                   y = -log10(q.val)))+  # 移除全局color映射
  geom_point(aes(color = type)) +  # 把color映射移到这里
  geom_text(data = subset(control, genename == "a"),
            aes(label = genename, vjust = 2)) +
  geom_point(data = treatment, aes(color = type)) +  # 这里也要加上color映射
  geom_text(data = subset(treatment, genename == "a"),
            aes(label = genename, vjust = 2)) +
  scale_color_manual(values = c("con" = "blue",
                                "treat" = "red"))

两种方法都能让图例恢复成单纯的圆形点，你可以根据自己的习惯选择。

内容的提问来源于stack exchange，提问作者user3138373