如何在R中组合宽度差异极大的plotly热力图子图？

首先，你的问题核心在于Plotly的subplot布局对极小宽度（<0.015）的子图存在渲染限制——它需要预留空间来渲染轴元素（刻度、标签等），当宽度过小的时候，这些元素的空间需求会挤压布局，导致显示错乱。下面提供两种更优的解决方案，既保留真实宽度比例，又能正常渲染：

方案一：优化Plotly子图的轴元素，适配极小宽度 #

我们可以针对极小宽度的子图，隐藏所有不必要的轴元素（刻度、标签、轴线等），让Plotly不需要为这些元素预留空间，从而正确适配极小的宽度比例。

修改后的代码： #

library(dplyr)
library(plotly)
library(grDevices)

set.seed(1)
df <- data.frame(row = rep(paste0("rid",1:100),10), 
                 col = paste0("cid",unlist(lapply(1:10,function(x) rep(x,100)))), 
                 val = rnorm(1000,-2,1))

plot.widths <- c(0.33277,0.0663,0.28308,0.09323,0.12969,0.0603,0.00651,0.01149,0.01503,0.0016)

# 生成绘图列表时，针对极小宽度的子图隐藏轴元素
plot.list <- lapply(1:10, function(i) {
  p <- plot_ly(z = c(filter(df, col == paste0("cid",i))$val),
               x = filter(df, col == paste0("cid",i))$col,
               y = filter(df, col == paste0("cid",i))$row,
               colors = colorRamp(c("darkblue","lightgray","darkred")),
               type = "heatmap") %>%
    layout(yaxis = list(title = NULL),
           xaxis = list(tickvals = i, ticktext = as.character(i)))
  
  # 如果当前子图宽度小于0.015，隐藏x轴的所有可见元素
  if (plot.widths[i] < 0.015) {
    p <- p %>% layout(xaxis = list(tickvals = NULL, ticktext = NULL, 
                                   showgrid = FALSE, showline = FALSE,
                                   zeroline = FALSE, visible = FALSE))
  }
  return(p)
})

# 组合子图，调整margin为极小值
subplot(plot.list, shareX = TRUE, shareY = TRUE, nrows = 1, margin = 0, widths = plot.widths) %>%
  layout(showlegend = FALSE)

原理： #

通过隐藏极小宽度子图的x轴元素，Plotly不再需要为这些元素分配空间，就能严格按照你设置的plot.widths比例渲染子图，同时共享Y轴保证整体一致性。

方案二：转为ggplot + gridExtra/grid布局（更稳定的精确布局） #

如果Plotly的方案还是有问题，完全可以转为用ggplot绘制热力图，再用gridExtra的grid.arrange来实现精确的宽度控制——grid系统对极小宽度的布局支持更好，不会因为轴元素挤压导致比例扭曲。

代码示例： #

library(dplyr)
library(ggplot2)
library(gridExtra)
library(grDevices)

set.seed(1)
df <- data.frame(row = rep(paste0("rid",1:100),10), 
                 col = paste0("cid",unlist(lapply(1:10,function(x) rep(x,100)))), 
                 val = rnorm(1000,-2,1))

plot.widths <- c(0.33277,0.0663,0.28308,0.09323,0.12969,0.0603,0.00651,0.01149,0.01503,0.0016)

# 生成ggplot热力图列表
ggplot.list <- lapply(1:10, function(i) {
  sub_df <- filter(df, col == paste0("cid",i))
  p <- ggplot(sub_df, aes(x = col, y = row, fill = val)) +
    geom_tile() +
    scale_fill_gradientn(colors = c("darkblue","lightgray","darkred")) +
    theme_minimal() +
    theme(axis.title.y = element_blank(),
          axis.text.y = ifelse(i == 1, element_text(), element_blank()), # 只保留第一个子图的Y轴标签
          axis.ticks.y = ifelse(i == 1, element_line(), element_blank()),
          axis.title.x = element_blank(),
          axis.text.x = ifelse(plot.widths[i] >= 0.015, element_text(as.character(i)), element_blank()),
          panel.grid = element_blank(),
          legend.position = "none")
  return(p)
})

# 用grid.arrange组合，设置精确宽度比例
grid.arrange(grobs = ggplot.list, nrow = 1, widths = plot.widths)

优势： #

grid系统的布局是基于精确的相对宽度/高度比例，不会像Plotly那样因为渲染元素的空间需求扭曲比例，完全保留你预设的宽度比例。同时通过条件控制只显示必要的轴标签，避免极小宽度子图的标签挤压。

内容的提问来源于stack exchange，提问作者dan