根据不同输入选择的分组数量动态调整绘图高度

需要根据选中的V1类别对应数据集中Cancer列的变量数量，动态调整ggplot绘图的高度。尝试通过layer_data(p, 1)获取分组数，再用length(table(out$group))*50计算高度，但无论V1选择多少类别，所有绘图高度始终一致。相关代码如下：

server <- function(input, output, session) {
  
  data_selected <- reactive({
    filter(files.Vir.DNA.df.test, V1 %in% input$Taxa)
  })
  
  output$myplot1 <- renderPlot({
    #data_selected() %>
    p <- ggplot(data_selected(),aes(position,rowSums, fill = Cancer)) + 
      geom_bar(stat="identity") +
      facet_grid(Cancer~. , scales = "free_x", space = "free_x", switch = "x") +
      theme(strip.text.y = element_text(angle = 0),
            strip.text.x = element_text(angle = 90),
            strip.background = element_rect(colour = "transparent", fill = "transparent"),
            plot.background = element_rect(colour = "white", fill = "white"),
            panel.background = element_rect(colour = "white", fill = "white"),
            
            axis.text.x = element_blank(),
            axis.ticks.x = element_blank()) + 
      labs(y="", x="", title="") +
      scale_fill_manual(values=mycolors) + 
      theme(legend.position = "none") +
      scale_y_log10(breaks=c(1,100,10000)) 
      print(p)
      out <- layer_data(p, 1)
    
  },res = 100,width = 600, height = length(table(out$group))*50)
}

1. 变量作用域问题：out <- layer_data(p, 1)定义在renderPlot的绘图逻辑内部，但height参数是renderPlot的外部参数。计算height时，out变量还未被当前绘图逻辑赋值，使用的是旧值或初始空值，导致高度无法动态更新。
2. 分组匹配误差：layer_data提取的group不一定完全对应Cancer的类别数，ggplot的图层分组与facet的分组逻辑可能存在差异，统计结果不准确。

方案1：直接从响应式数据统计类别数（推荐） #

无需从绘图对象提取信息，直接从data_selected()中统计Cancer的唯一值数量，将高度计算设为响应式表达式，确保数据更新时高度同步调整：

server <- function(input, output, session) {
  
  data_selected <- reactive({
    filter(files.Vir.DNA.df.test, V1 %in% input$Taxa)
  })
  
  # 响应式计算Cancer的类别数量
  cancer_category_count <- reactive({
    length(unique(data_selected()$Cancer))
  })
  
  output$myplot1 <- renderPlot({
    p <- ggplot(data_selected(), aes(position, rowSums, fill = Cancer)) + 
      geom_bar(stat="identity") +
      facet_grid(Cancer~. , scales = "free_x", space = "free_x", switch = "x") +
      theme(strip.text.y = element_text(angle = 0),
            strip.text.x = element_text(angle = 90),
            strip.background = element_rect(colour = "transparent", fill = "transparent"),
            plot.background = element_rect(colour = "white", fill = "white"),
            panel.background = element_rect(colour = "white", fill = "white"),
            axis.text.x = element_blank(),
            axis.ticks.x = element_blank()) + 
      labs(y="", x="", title="") +
      scale_fill_manual(values=mycolors) + 
      theme(legend.position = "none") +
      scale_y_log10(breaks=c(1,100,10000)) 
    print(p)
  }, res = 100, width = 600, height = reactive(cancer_category_count() * 50)())
}

方案2：从绘图对象提取分组信息 #

如果需要从绘图对象中获取分组数据，可以使用ggplot_build构建完整绘图对象后提取facet分组数，同时确保高度计算为响应式：

server <- function(input, output, session) {
  
  data_selected <- reactive({
    filter(files.Vir.DNA.df.test, V1 %in% input$Taxa)
  })
  
  output$myplot1 <- renderPlot({
    req(data_selected())
    p <- ggplot(data_selected(), aes(position, rowSums, fill = Cancer)) + 
      geom_bar(stat="identity") +
      facet_grid(Cancer~. , scales = "free_x", space = "free_x", switch = "x") +
      theme(strip.text.y = element_text(angle = 0),
            strip.text.x = element_text(angle = 90),
            strip.background = element_rect(colour = "transparent", fill = "transparent"),
            plot.background = element_rect(colour = "white", fill = "white"),
            panel.background = element_rect(colour = "white", fill = "white"),
            axis.text.x = element_blank(),
            axis.ticks.x = element_blank()) + 
      labs(y="", x="", title="") +
      scale_fill_manual(values=mycolors) + 
      theme(legend.position = "none") +
      scale_y_log10(breaks=c(1,100,10000)) 
    print(p)
  }, res = 100, width = 600, height = reactive({
    req(data_selected())
    # 构建绘图对象并提取Cancer的分组数
    plot_obj <- ggplot_build(ggplot(data_selected(), aes(position, rowSums, fill = Cancer)) + 
                               geom_bar(stat="identity") +
                               facet_grid(Cancer~. , scales = "free_x", space = "free_x", switch = "x"))
    length(unique(plot_obj$layout$panel_layout$Cancer)) * 50
  })())
}

核心要点 #

• renderPlot的height参数必须是响应式表达式，这样当input$Taxa变化导致data_selected()更新时，高度会自动重新计算。
• 直接从原始响应式数据统计Cancer类别数，比从绘图对象提取更可靠，避免分组逻辑不匹配的问题。

内容的提问来源于stack exchange，提问作者user2300940