用户为ggplot2 geom_boxplot定义参数

I have the following data.frame:

我有以下数据。

df = data.frame(ymin = c(0.35,0.4,0.25,0.3,0.55,0.6), lower = c(0.45,0.5,0.35,0.4,0.65,0.7), middle = c(0.5,0.55,0.4,0.45,0.7,0.75), upper = c(0.55,0.6,0.45,0.5,0.75,0.8), ymax = c(0.65,0.7,0.55,0.6,0.85,0.9), factor = c("parental","parental","cross","cross","sex","sex"), factor.label = c("paternal","maternal","F1i","F1R","M","F"), posterior.probability = c(0.92,0.92,0.97,0.97,0.99,0.99), x = c(1,1,2,2,3,3), colors = c("blue","red","gray30","gray70","lightskyblue","pink"))

I want to produce a ggplot2 box plot where df$x defines the x axis locations of the boxes and df$lower, df$upper, df$middle, df$ymin, df$ymax define the boxes, and df$colors define the colors of the boxes. As you can see each pair of boxes should be put on the same x axis location. With the alpha parameter I'll make the boxes transparent so overlaps will be visible.

我想制作一个ggplot2 box plot，其中df$x定义了盒子的x轴位置，df$lower, df$upper, df$middle, df$ymin, df$ymax定义了盒子的颜色，df$彩色定义了盒子的颜色。正如您所看到的，每一对盒子都应该放在相同的x轴位置上。有了alpha参数，我将使盒子透明，这样重叠就可以看到了。

What I have so far is this code:

到目前为止，我的代码是:

p = ggplot(beta.df, aes(x = x, color = colors))
p = p + geom_boxplot(aes(lower = lower,upper = upper, middle = middle, ymin = ymin, ymax = ymax, fill = colors), position = position_dodge(width = 0), width = 0.5, alpha = 0.5, stat = "identity")

Obviously the colors of the boxes are messed up - the "blue" and "red" boxes should be at df$x = 1, the "gray30" and "gray70" boxes should be at df$x = 2, and , the "lightskyblue" and "pink" boxes should be at df$x = 3. So I'm looking to fix that.

很明显，盒子的颜色弄乱了——“蓝色”和“红色”的盒子应该是df$x = 1，“gray30”和“gray70”的盒子应该是df$x = 2，“lightskyblue”和“pink”的盒子应该是df$x = 3。所以我想解决这个问题。

I additionally want to have the legend title and labels to be specified rather then generated by defaults. In addition, the df$posterior.probability values are identical for each pair of boxes with the same x axis location, and what I would like is to draw these values (one for each pair) either at the top of the plot, say at the maximum y axis value, or on top the max(ymax) of each pair of boxes and at the corresponding x axis locations. In other words, 0.92, 0.97, and 0.99 will be drawn at x locations: 1, 2, and 3, respectively, and at either the maximum y location of the plot or at these locations y locations: 0.75, 0.65, and 0.95, respectively.

另外，我还希望将标题和标签指定为默认值，而不是由默认值生成。此外,df后美元。概率值是相同的每一对盒相同的x轴的位置,我想画这些值(每组一个)顶部的阴谋,在y轴的最大价值,或者在上面的马克斯(ymax)每一对对应的盒子和x轴的位置。换句话说，0.92、0.97和0.99将分别绘制在x个地点:1、2和3，以及在地块的最大y位置，或分别绘制在这些地点:0.75、0.65和0.95。

2 个解决方案

#1

Your boxplots are overlapping and as I understand from your code, you don't want that. You don't need the position element in your code and you have tot treat you x as a factor in order to get them plotted next to each other.

你的boxplot是重叠的，我从你的代码中了解到，你不希望这样。你不需要在你的代码中使用位置元素，你必须把你的x作为一个因数来把它们画在一起。

With this code:

这段代码:

ggplot(df, aes(x = as.factor(x), color = colors)) + 
  geom_boxplot(aes(lower = lower,upper = upper, middle = middle, ymin = ymin, ymax = ymax, fill = colors), width = 0.5, alpha = 0.5, stat = "identity") +
  geom_text(data = df, aes(x = as.factor(x), y = ymax, label = posterior.probability), size = 4, vjust = -0.5) +
  scale_fill_identity("Color legend", guide = "legend", labels = c("paternal","F1i","F1R","M","F","maternal")) +
  scale_color_identity("Color legend", guide = "legend", labels = c("paternal","F1i","F1R","M","F","maternal")) +
  labs(title = "Plot title", x = "X-lab label", y = "Y-lab label")

You get this result:

你得到这个结果:

When you don't want the text-labels in the same color, add color = "black" to the geom_text part.

如果不希望文本标签使用相同的颜色，请在geom_text部分中添加color = "black"。

#2

For labels made new data frame that contains only one label for each x position and y value is calculated as maximal ymax value.

对于标签，新数据帧只包含每个x位置的一个标签，y值被计算为最大ymax值。

library(plyr)
df.text<-ddply(df,.(x),summarise,y=max(ymax),label=max(posterior.probability))

As you supply color names then you should use scale_fill_identity() and scale_color_identity() to tell ggplot to interpret those as actual color. If you need to show legend, then add argument guide="legend" to scale_fill_identity() and scale_color_identity() and then provide labels= you want to show in legend. Then use geom_text() and new data frame to add labels above boxplots.

当您提供颜色名称时，您应该使用scale_fill_identity()和scale_color_identity()来告诉ggplot将它们解释为实际的颜色。如果需要显示legend，那么向scale_fill_identity()和scale_color_identity()添加参数指南=“legend”，然后提供要在legend中显示的标签=。然后使用geom_text()和新的数据帧在boxplot上添加标签。

ggplot(df, aes(x = x)) + 
  geom_boxplot(aes(lower = lower,upper = upper, middle = middle, ymin = ymin, 
                    ymax = ymax,color = colors, fill = colors), 
                    position = position_dodge(width = 0), width = 0.5, 
                     alpha = 0.5, stat = "identity")+
  scale_fill_identity("Legend name",guide="legend",
                      labels=c("paternal","F1i","F1R","M","F","maternal"))+
    scale_color_identity("Legend name",guide="legend",
                      labels=c("paternal","F1i","F1R","M","F","maternal"))+
  geom_text(data=df.text,aes(x=x,y=y,label=label))