是否可以创建一个具有两种颜色之间渐变的geom_segment，但开始/结束颜色根据变量而变化？

Question

我正在构建一个具有离散 Y 轴的图，并进行多个成对比较。然而，我没有使用文本作为标签，而是创建了一个辅助图，其中的颜色点指示参与比较的 2 个组，为了清晰起见，用一个线段连接两个点。见左图：

我用来生成左侧图的代码是：

p_left <- ggplot(tb_deg) +
  geom_segment(aes(y=comparison, x=var1, xend=var2), size=2, color="slategrey", show.legend = F) +
  geom_point(aes(y=comparison, x=var1, fill=var1), shape=21, size=5, show.legend = F) +
  geom_point(aes(y=comparison, x=var2, fill=var2), shape=21, size=5, show.legend = F) +
  scale_fill_manual(name="Group",
                    values=c('ControlD1' = "limegreen", 
                             'HSP90iD1' = "firebrick1", 
                             'PRRTD1' = "deepskyblue", 
                             'CombinationD1' = "orchid1", 
                             'HSP90iD3' = "firebrick4", 
                             'PRRTD3' = "deepskyblue3", 
                             'CombinationD3' = "purple3")
                    ) +
  scale_x_discrete(limits=rev(levels(md$group)) ) +
  theme_void() +
  labs(x=NULL, y=NULL, title = NULL) +
  theme(legend.position = "none",
        aspect.ratio = 2,
        axis.text.x = element_text(angle=90, hjust=1),
        axis.text.y = element_blank())

现在我正在将连接线段设置为灰色，但我想知道是否可以将该线段设置为两个连接点的颜色之间的渐变。是的，它可能会很难看，我不会在最终的图中使用它，但我想尝试一下。

我一直在寻找如何做到这一点，但我发现的所有解决方案都使用固定（渐变）色标，该色标对于所有绘制的段都相同（主要使用 geom_link）。我找不到任何片段尖端颜色取决于变量的示例。理想情况下，我会寻找类似的东西

    ... + geom_segment(aes(y=comparison, x=var1, xend=var2, 
***color_start=var1, color_end=var2***) ) + ...

有什么方法可以做到这一点，而无需手动为 15 次比较中的每一个比较编码具有不同比例的渐变吗？

谢谢，

编辑：数据：

> dput(unique(tb_deg[,c("comparison","var1","var2")]))
structure(list(comparison = structure(c(3L, 7L, 8L, 15L, 6L, 
10L, 11L, 1L, 4L, 13L, 2L, 9L, 5L, 12L, 14L), levels = c("without001 - HSP90iD1_vs_ControlD1", 
"without001 - PRRTD1_vs_ControlD1", "without001 - CombinationD1_vs_ControlD1", 
"without001 - HSP90iD3_vs_ControlD1", "without001 - PRRTD3_vs_ControlD1", 
"without001 - CombinationD3_vs_ControlD1", "without001 - CombinationD1_vs_HSP90iD1", 
"without001 - CombinationD1_vs_PRRTD1", "without001 - PRRTD1_vs_HSP90iD1", 
"without001 - CombinationD3_vs_HSP90iD3", "without001 - CombinationD3_vs_PRRTD3", 
"without001 - PRRTD3_vs_HSP90iD3", "without001 - HSP90iD3_vs_HSP90iD1", 
"without001 - PRRTD3_vs_PRRTD1", "without001 - CombinationD3_vs_CombinationD1"
), class = "factor"), var1 = structure(c(4L, 4L, 4L, 7L, 7L, 
7L, 7L, 2L, 5L, 5L, 3L, 3L, 6L, 6L, 6L), levels = c("ControlD1", 
"HSP90iD1", "PRRTD1", "CombinationD1", "HSP90iD3", "PRRTD3", 
"CombinationD3"), class = "factor"), var2 = structure(c(1L, 2L, 
3L, 4L, 1L, 5L, 6L, 1L, 1L, 2L, 1L, 2L, 1L, 5L, 3L), levels = c("ControlD1", 
"HSP90iD1", "PRRTD1", "CombinationD1", "HSP90iD3", "PRRTD3", 
"CombinationD3"), class = "factor")), class = c("grouped_df", 
"tbl_df", "tbl", "data.frame"), row.names = c(NA, -15L), groups = structure(list(
    comparison = structure(1:15, levels = c("without001 - HSP90iD1_vs_ControlD1", 
    "without001 - PRRTD1_vs_ControlD1", "without001 - CombinationD1_vs_ControlD1", 
    "without001 - HSP90iD3_vs_ControlD1", "without001 - PRRTD3_vs_ControlD1", 
    "without001 - CombinationD3_vs_ControlD1", "without001 - CombinationD1_vs_HSP90iD1", 
    "without001 - CombinationD1_vs_PRRTD1", "without001 - PRRTD1_vs_HSP90iD1", 
    "without001 - CombinationD3_vs_HSP90iD3", "without001 - CombinationD3_vs_PRRTD3", 
    "without001 - PRRTD3_vs_HSP90iD3", "without001 - HSP90iD3_vs_HSP90iD1", 
    "without001 - PRRTD3_vs_PRRTD1", "without001 - CombinationD3_vs_CombinationD1"
    ), class = "factor"), var1 = structure(c(2L, 3L, 4L, 5L, 
    6L, 7L, 4L, 4L, 3L, 7L, 7L, 6L, 5L, 6L, 7L), levels = c("ControlD1", 
    "HSP90iD1", "PRRTD1", "CombinationD1", "HSP90iD3", "PRRTD3", 
    "CombinationD3"), class = "factor"), .rows = structure(list(
        8L, 11L, 1L, 9L, 13L, 5L, 2L, 3L, 12L, 6L, 7L, 14L, 10L, 
        15L, 4L), ptype = integer(0), class = c("vctrs_list_of", 
    "vctrs_vctr", "list"))), class = c("tbl_df", "tbl", "data.frame"
), row.names = c(NA, -15L), .drop = TRUE))

Answer 1

同时使用 ggforce 的 geom_link 和 ggnewscale，可以逐个元素创建一个循环，为每个元素创建一个新的色阶：

library(ggforce)
library(ggnewscale)

## Index with colors
color_list <- c('ControlD1' = "limegreen", 
                             'HSP90iD1' = "firebrick1", 
                             'PRRTD1' = "deepskyblue", 
                             'CombinationD1' = "orchid1", 
                             'HSP90iD3' = "firebrick4", 
                             'PRRTD3' = "deepskyblue3", 
                             'CombinationD3' = "purple3")

## old-ish code
p_left <- ggplot(tb_deg) +
  geom_segment(aes(y=comparison, x=var1, xend=var2), size=2, color="slategrey", show.legend = F) +
  geom_point(aes(y=comparison, x=var1, fill=var1), shape=21, size=5, show.legend = F) +
  geom_point(aes(y=comparison, x=var2, fill=var2), shape=21, size=5, show.legend = F) +
  scale_fill_manual(name="Group",
                    values=color_list ) +
  scale_x_discrete(limits=rev(levels(md$group)) ) +
  theme_minimal() +
  labs(x=NULL, y=NULL, title = NULL) +
  theme(legend.position = "none",
        aspect.ratio = 2,
        axis.text.x = element_text(angle=90, hjust=1),
        axis.text.y = element_blank(),
        panel.grid.major.y = element_blank())



p_left

### Loop through each element of the Y-axis, creating a geom_link + a new color scale
p2<-p_left

for (comp in unique(tb_deg$comparison)) {

  v1 <-   unique(tb_deg[tb_deg$comparison==comp, c("comparison","var1","var2")])$var1
  v2 <-   unique(tb_deg[tb_deg$comparison==comp, c("comparison","var1","var2")])$var2

p2 <- p2 +
  new_scale_color() +
  geom_link(data=unique(tb_deg[tb_deg$comparison==comp, c("comparison","var1","var2")]),
            aes(y=comparison, yend=comparison, x=var1, xend=var2, color=stat(index)), size=2) +
  scale_color_gradient(high=color_list[v2],low=color_list[v1])
  
}

p2