R语言cowplot包axis_canvas报错：x must be length 1 or 2

我用ggplot绘制散点图并尝试添加边际箱线图，因ggmarginal生成的箱线图宽度不一致，选择使用cowplot包。x轴边际箱线图可正常生成，但添加y轴边际箱线图时触发错误：Error in zero_range(range) : x must be length 1 or 2。

散点图代码 #

data_merged %>%  
  filter(treatment != "ctr") %>% 
  split(.$compartment)  %>% 
  map(
    ~ggplot(., 
            aes(x=time_after_treatment, y=max_peak, 
                fill=treatment, colour=treatment))+
      scale_colour_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3")) +
      scale_fill_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3")) +
      geom_point(size=3) +
      scale_x_continuous(limits= c(2.5,17))+
      scale_y_continuous (limits= c(0.4, 2.0, 0.5)) +  # 错误根源
      theme_classic() +
      labs(y= Ca^"2+"~peak~(mu*M), x= "Time (min)") +
      theme(#axis.title.x=element_blank(), 
            legend.position = "none",
            axis.text=element_text(size=11),
            panel.border = element_blank(),
            panel.grid.minor = element_blank(),
            strip.background = element_blank()) 
  ) -> list_scatter

正常运行的x轴边际箱线图代码 #

data_merged %>%  
  filter(treatment != "ctr") %>% 
  filter(compartment == "cytosol") -> data_cyt 

xbox_P3 <- axis_canvas(list_scatter[[1]], axis = "x", coord_flip = TRUE) +
  geom_boxplot(data=data_cyt, 
               aes(y=time_after_treatment, x=treatment, fill=treatment))+
  scale_fill_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3"))+
  scale_x_discrete() +coord_flip()

P3 <- insert_xaxis_grob(list_scatter[[1]], xbox_P3, 
                        grid::unit(0.9, "in"), position="top")
P3_plot <- ggdraw(P3)

报错的y轴边际箱线图代码 #

ybox_P4 <- axis_canvas(list_scatter[[2]], axis = "y") +
  geom_boxplot(data=data_cyt, 
               aes(y=max_peak, x=treatment, fill=treatment))+
  scale_fill_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3"))+
  scale_x_discrete()

错误信息 #

Error in zero_range(range) : x must be length 1 or 2
5. stop("x must be length 1 or 2")
4. zero_range(range)
3. scales::expand_range(range, expand[1], expand[2])
2. get_scale_limits(layer_scales(plot)$y)

1. axis_canvas(list_scatter[[2]], axis = "y")

错误原因 #

问题出在散点图代码的scale_y_continuous参数上：limits= c(0.4, 2.0, 0.5)。limits参数只需要传入长度为2的向量（最小值和最大值），但额外添加的第三个值0.5导致y轴范围解析异常。当axis_canvas尝试读取y轴范围时，调用的zero_range函数要求输入的范围向量长度必须是1或2，因此触发错误。

解决方法 #

将scale_y_continuous的limits参数修正为仅包含最小值和最大值，若需要设置刻度间隔，改用breaks参数：

修正后的散点图代码片段 #

scale_y_continuous(limits= c(0.4, 2.0), breaks = seq(0.4, 2.0, 0.5)) +

完整修正后的散点图代码 #

data_merged %>%  
  filter(treatment != "ctr") %>% 
  split(.$compartment)  %>% 
  map(
    ~ggplot(., 
            aes(x=time_after_treatment, y=max_peak, 
                fill=treatment, colour=treatment))+
      scale_colour_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3")) +
      scale_fill_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3")) +
      geom_point(size=3) +
      scale_x_continuous(limits= c(2.5,17))+
      scale_y_continuous(limits= c(0.4, 2.0), breaks = seq(0.4, 2.0, 0.5)) +
      theme_classic() +
      labs(y= Ca^"2+"~peak~(mu*M), x= "Time (min)") +
      theme(#axis.title.x=element_blank(), 
            legend.position = "none",
            axis.text=element_text(size=11),
            panel.border = element_blank(),
            panel.grid.minor = element_blank(),
            strip.background = element_blank()) 
  ) -> list_scatter

修正后重新运行y轴箱线图代码 #

此时再执行报错的y轴箱线图代码，即可正常生成边际箱线图：

ybox_P4 <- axis_canvas(list_scatter[[2]], axis = "y") +
  geom_boxplot(data=data_cyt, 
               aes(y=max_peak, x=treatment, fill=treatment))+
  scale_fill_manual(values=c("#1f78b4", "#b2df8a", "#a6cee3"))+
  scale_x_discrete()

# 插入到散点图中
P4 <- insert_yaxis_grob(list_scatter[[2]], ybox_P4, 
                        grid::unit(0.9, "in"), position="right")
P4_plot <- ggdraw(P4)

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