在ggplot2中插入图例下的表

Question

我发布了这个问题earlier，但事实证明我确认的答案有问题，所以我再次发布这个问题，希望能解决这个问题。

我有下面的data.frame，我想用ggplot2绘制成一个文件：

df = data.frame(mean=c(1.96535,2.133604,1.99303,1.865004,2.181713,1.909511,2.047971,1.676599,2.143763,1.939875,1.816028,1.95465,2.153445,1.802517,2.141799,1.722428),
sd=c(0.0595173,0.03884202,0.0570006,0.04934336,0.04008221,0.05108064,0.0463556,0.06272475,0.04321496,0.05283728,0.05894342,0.05160038,0.04679423,0.05305525,0.04626291,0.0573123),
par=as.factor(c("p","p","m","m","p","p","m","m","m","m","p","p","m","m","p","p")),
group=as.factor(c("iF","iF","iF","iF","iM","iM","iM","iM","RF","RF","RF","RF","RM","RM","RM","RM")),
rep=c(1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2)) 

这是我的代码，用于生成ggplot2对象：

p <- ggplot(df,aes(factor(rep),y=mean,ymin=mean-2*sd,ymax=mean+2*sd,color=factor(par)))
p <- p + geom_pointrange()+facet_wrap(~group, ncol = 4)+scale_color_manual(values = c("blue","red"),labels = c("p","m"),name = "par id")
p <- p + ggtitle("test")
p <- p + labs(y="log(y)",x="rep")

另外，我要做的是将这个data.frame作为一个表添加到图例中：

leg.df = data.frame(statistic = c("pp(par)","pp(g)","pp(s)","fc(p/m)"), value = c(0.96,0.94,0.78,1.5))

我得到了这样的解决方案：

leg.df.grob <-  tableGrob(leg.df, gpar.coretext =gpar(fontsize=8),
         par.coltext=gpar(fontsize=8), 
         gpar.rowtext=gpar(fontsize=8))

### final result
library(gridExtra)
pp <- arrangeGrob(p + theme(legend.position = "none"), 
                  arrangeGrob(leg.df.grob, legend), ncol = 2)

然后，理论上，我可以将pp保存到一个文件中，例如，使用：

ggsave('plot.png',pp)

不幸的是，命令：

pp <- arrangeGrob(p + theme(legend.position = "none"), 
                      arrangeGrob(leg.df.grob, legend), ncol = 2) 

导致绘图设备打开(这正是我试图避免的)，此外，它还抛出了以下错误：

Error in arrangeGrob(leg.df.grob, legend) : input must be grobs!

知道怎么解决这个问题吗？

Answer 1

另一个解决方案是正确的。我想您得到了一个错误，因为您没有设置legend变量。因此，以R函数legend作为参数调用legend。您应该将legend定义为：

g_legend <- function(a.gplot){
    tmp <- ggplot_gtable(ggplot_build(a.gplot))
    leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
    legend <- tmp$grobs[[leg]]
    return(legend)
}
legend <- g_legend(p)

我稍微修改了另一个答案，通过设置widths参数来更好地重新排列grobs：

pp <- arrangeGrob(p + theme(legend.position = "none"), 
                  widths=c(3/4, 1/4),
                  arrangeGrob( legend,leg.df.grob), ncol = 2)

​

​