在 plotly 中复制带有刻面和箭头注释的 ggplot2 图表

Question

我想在 plotly 中复制一个 ggplot2 图表，为此我尝试使用 ggplotly 函数。 图表如下 (p_arrow):

library(data.table)
library(ggplot2)
library(plotly)

dset <- data.table(x = c(0, 1),
                   y = c(0, 1),
                   treatment = c(0, 1))

x_start <- dset[treatment == 0, x]
x_end <- dset[treatment == 1, x]
y_start <- dset[treatment == 0, y]
y_end <- dset[treatment == 1, y]
# base chart
p <- ggplot(dset, aes(x, y)) + geom_point() + facet_grid( ~ treatment)
# add annotations
p_arrow <- p +   annotate(
  "segment",
  x = x_start,
  xend = x_end,
  y = y_start,
  yend = y_end,
  lty = 1,
  size = 1,
  arrow = arrow()
)
# chart to be replicated
p_arrow 
# If I convert to plotly I lose the arrow
ggplotly(p_arrow)

plotly 版本没有箭头。这是 know issue因此，我一直在寻找一种直接在 plotly 中添加注释的方法。

第一次试验:

# my trials in ggplotly
pl <- ggplotly(p)
#pl1 only annotates the fist subplot
pl1 <- pl %>% add_annotations(
  x = x_start,
  y = y_start,
  xref = "x",
  yref = "y",
  axref = "x",
  ayref = "y",
  text = "",
  showarrow = TRUE,
  ax = x_end,
  ay = y_end,
  arrowcolor = "black"
)

pl1

问题:第二个子图没有注释。

二审:

# add annotation in axis x2 and y2
pl2 <- pl1 %>% add_annotations(
  x = x_start,
  y = y_start,
  xref = "x2",
  yref = "y2",
  axref = "x2",
  ayref = "y2",
  text = "",
  showarrow = TRUE,
  ax = x_end,
  ay = y_end,
  arrowcolor = "black"
)

pl2

问题:第二个子图中缺少点。

有没有办法复制 p_arrow？

Answer 1

如果在对 add_annotations 的两次调用中将 yref 设置为同一轴(即 "y" 而不是 "y2")，那么问题就解决了:

pl <- ggplotly(p)

pl <- pl %>% add_annotations(
        x = x_end,
        y = y_end,
        xref = "x2",
        yref = "y",
        axref = "x2",
        ayref = "y",
        text = "",
        showarrow = TRUE,
        arrowhead = 3,
        arrowsize = 2,
        ax = x_start,
        ay = y_start
) %>% add_annotations(
        x = x_end,
        y = y_end,
        xref = "x",
        yref = "y",
        axref = "x",
        ayref = "y",
        text = "",
        showarrow = TRUE,
        arrowhead = 3,
        arrowsize = 2,
        ax = x_start,
        ay = y_start
)

所以现在 pl 看起来像这样:

与原始版本相似(实际上比原始版本更好):