PCA ggplot绘图中能否将变量箭头替换为仅显示点和标签

可以实现，只需调整ggbiplot的内置参数即可，操作方案如下：

方案1：仅隐藏变量箭头，保留样本点与分组信息 #

在ggbiplot函数中添加参数var.axes = FALSE即可关闭变量箭头绘制，修改后代码如下：

biplot <- ggbiplot(pcobj = PCA, choices = c(1,2), scale = 1, 
                   groups = Fdata$Tissue, ellipse = TRUE,
                   var.axes = FALSE) + 
  geom_vline(xintercept = 0, linetype = 3) +
  geom_hline(yintercept = 0, linetype = 3) + 
  geom_point(aes(shape = factor(Fdata$Tissue))) + 
  guides(color = "none", shape = guide_legend(title = "Tissue")) + 
  scale_shape_discrete(labels = c("Flesh","Gill"))

方案2：隐藏箭头同时展示变量的点与对应标签 #

如果需要在原箭头位置显示变量对应的点和名称标签，额外添加变量点、标签控制参数即可，修改后代码如下：

biplot <- ggbiplot(pcobj = PCA, choices = c(1,2), scale = 1, 
                   groups = Fdata$Tissue, ellipse = TRUE,
                   var.axes = FALSE, 
                   var.point.size = 3, 
                   var.label = TRUE, 
                   var.label.size = 3.5) + 
  geom_vline(xintercept = 0, linetype = 3) +
  geom_hline(yintercept = 0, linetype = 3) + 
  geom_point(aes(shape = factor(Fdata$Tissue))) + 
  guides(color = "none", shape = guide_legend(title = "Tissue")) + 
  scale_shape_discrete(labels = c("Flesh","Gill"))

参数说明 #

• var.axes默认值为TRUE，控制是否绘制变量箭头，设为FALSE后即可隐藏箭头
• 可根据可视化需求调整var.point.size（变量点大小）、var.label.size（变量标签字体大小）的数值

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