ggplotly转换ggplot2的geom_point图时填充渐变丢失的解决办法

问题现象 #

使用ggplot2的geom_point(shape=21)创建散点图时，通过fill映射实现填充色渐变、color映射区分显著性、size映射区分分数，图形显示正常。但经plotly::ggplotly()转换后，填充渐变完全消失，所有点填充色一致，同时出现两条"替换长度不匹配"的警告信息。

问题原因 #

plotly对ggplot中shape=21这类带独立填充色和边框色的形状的映射转换存在兼容问题：ggplot的fill和color在转换时被plotly错误地合并处理，导致渐变填充的映射逻辑失效。

解决方案 #

通过预计算填充色与边框色，再手动指定plotly的marker属性，绕开ggplotly的转换bug。以下提供两种可行实现方式：

方法一：预计算颜色后修正ggplotly对象 #

先在数据中生成对应level的渐变填充色和sig的边框色，再将ggplot转换为plotly后手动覆盖marker属性：

library(dplyr)
library(ggplot2)
library(plotly)
library(scales)
library(reshape2)

# 生成原始数据
set.seed(1)
levels.mat <- matrix(runif(7*6, 1, 3), nrow = 7, ncol = 6, 
                     dimnames = list(paste0("I", 1:7), paste0("P", 1:6)))
scores.mat <- matrix(runif(7*6, 0, 10), nrow = 7, ncol = 6, 
                     dimnames = list(paste0("I", 1:7), paste0("P", 1:6)))
sig.mat <- matrix(sample(c(F, T), 7*6, replace = T), nrow = 7, ncol = 6, 
                  dimnames = list(paste0("I", 1:7), paste0("P", 1:6)))

df <- left_join(melt(levels.mat) %>% rename(x = Var1, y = Var2, level = value),
                melt(scores.mat) %>% rename(x = Var1, y = Var2, score = value)) %>%
  left_join(melt(sig.mat) %>% rename(x = Var1, y = Var2, sig = value))

# 预计算渐变填充色和边框色
df$fill_color <- gradient_n_pal(c("lightgray", "darkred"))(rescale(df$level))
df$border_color <- ifelse(df$sig, "green", "black")

# 创建基础ggplot（仅保留size映射，避免转换冲突）
p <- ggplot(df, aes(x = x, y = y)) +
  geom_point(aes(size = score), shape = 21, fill = NA, color = NA) +
  theme_minimal() +
  theme(axis.title = element_blank(),
        axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1))

# 转换为plotly并手动设置marker属性
p_plotly <- ggplotly(p) %>%
  style(
    marker = list(
      color = df$border_color,       # 设置边框色
      fillcolor = df$fill_color,     # 设置填充渐变
      sizemode = "area"              # 匹配ggplot的大小缩放逻辑
    )
  )

# 手动补充颜色图例（可选）
p_plotly <- p_plotly %>%
  colorbar(title = "Level", colors = c("lightgray", "darkred"), limits = range(df$level)) %>%
  add_annotations(xref = "paper", yref = "paper", x = 1.05, y = 1, 
                  text = "Significant", showarrow = F, font = list(color = "green")) %>%
  add_annotations(xref = "paper", yref = "paper", x = 1.05, y = 0.95, 
                  text = "Not Significant", showarrow = F, font = list(color = "black"))

# 输出图形
p_plotly

方法二：直接使用plotly原生语法构建图形 #

完全绕过ggplot，直接用plotly原生函数创建图形，避免转换兼容问题：

library(plotly)
library(scales)
library(reshape2)
library(dplyr)

# 生成原始数据（同方法一，省略重复代码）
# ...

# 预计算颜色（同方法一）
df$fill_color <- gradient_n_pal(c("lightgray", "darkred"))(rescale(df$level))
df$border_color <- ifelse(df$sig, "green", "black")

# 原生plotly绘图
plot_ly(df, x = ~x, y = ~y, type = "scatter", mode = "markers") %>%
  add_markers(
    size = ~score,
    marker = list(
      color = ~border_color,
      fillcolor = ~fill_color,
      line = list(width = 1),
      sizemode = "area",
      sizeref = 2 * max(df$score) / (40^2)  # 调整大小缩放，匹配ggplot视觉效果
    ),
    text = ~paste("Level:", round(level, 2), "<br>Score:", round(score, 2), "<br>Significant:", sig)
  ) %>%
  layout(
    xaxis = list(title = "", tickangle = 45),
    yaxis = list(title = ""),
    showlegend = FALSE,
    annotations = list(
      list(xref = "paper", yref = "paper", x = 1.05, y = 1, 
           text = "Significant", showarrow = F, font = list(color = "green")),
      list(xref = "paper", yref = "paper", x = 1.05, y = 0.95, 
           text = "Not Significant", showarrow = F, font = list(color = "black"))
    )
  ) %>%
  colorbar(title = "Level", colors = c("lightgray", "darkred"), limits = range(df$level))

关键说明 #

• 使用scales包的gradient_n_pal和rescale函数，确保预计算的渐变颜色与原ggplot的scale_fill_gradient逻辑完全一致
• sizemode = "area"是ggplot默认的点大小映射规则，必须设置才能和原图形视觉匹配
• 手动添加图例是因为修正ggplotly后，原ggplot的图例可能丢失，需按需补充

内容的提问来源于stack exchange，提问作者dan