R中cowplot包问题：拼接grob并导出PDF时项目符号变为椭圆

使用consort包生成流程图，单独绘制显示正常，但通过cowplot拼接后导出PDF，流程图中的项目符号会变成椭圆，导出PNG则无此问题。最小复现代码如下：

library(consort)
library(cowplot)
library(ggplotify)

# 生成流程图A
participant_id <- 1:10
exclusion <- rep(NA, 10)
exclusion[1:3] <- "Lost to follow-up"
df <- data.frame(participant_id, exclusion)
flowchart_a <- consort_plot(data = df,
                            orders = c(participant_id = "Invited participants",
                                       exclusion = "Excluded",
                                       participant_id = "Completed the study"),
                            side_box = c("exclusion"),
                            cex = 0.9)

# 生成流程图B
rm(participant_id, exclusion, df)
participant_id <- 1:10
exclusion <- rep(NA, 10)
exclusion[1:2] <- "Lost to follow-up"
df <- data.frame(participant_id, exclusion)
flowchart_b <- consort_plot(data = df,
                            orders = c(participant_id = "Invited participants",
                                       exclusion = "Excluded",
                                       participant_id = "Completed the study"),
                            side_box = c("exclusion"),
                            cex = 0.9)

# 拼接并导出PDF
grob1 <- as.grob(function() plot(flowchart_a))
grob2 <- as.grob(function() plot(flowchart_b))
grid <- plot_grid(grob1, NULL, grob2,
                  rel_heights = c(1, 0.3, 1),
                  labels = c("A", "", "B"),
                  ncol = 1)
save_plot("panel.pdf", grid, nrow = 2, ncol = 1.5)

原因分析 #

该问题源于consort包基于base R绘图系统生成流程图，其项目符号依赖base R的pch参数定义。当通过ggplotify::as.grob()将base绘图转为grob对象，再用cowplot::save_plot()导出PDF时，矢量格式下的符号渲染逻辑出现异常，导致圆形符号被拉伸为椭圆；而PNG是位图格式，渲染逻辑不受此影响。

解决方案 #

方法1：改用base R原生PDF设备保存 #

绕过cowplot::save_plot()，直接使用base R的pdf()设备输出拼接后的图形，避免矢量渲染异常：

# （生成流程图A、B及拼接代码同原示例）

# 使用base R的pdf设备保存
pdf("panel.pdf", width = 8, height = 10)
grid.draw(grid)
dev.off()

方法2：自定义consort流程图的项目符号 #

修改consort_plot()的pch参数，使用矢量友好的Unicode圆形符号，避免拉伸问题：

# 生成流程图时指定pch为Unicode实心圆
flowchart_a <- consort_plot(data = df,
                            orders = c(participant_id = "Invited participants",
                                       exclusion = "Excluded",
                                       participant_id = "Completed the study"),
                            side_box = c("exclusion"),
                            cex = 0.9,
                            pch = "\u25CF")

flowchart_b <- consort_plot(data = df,
                            orders = c(participant_id = "Invited participants",
                                       exclusion = "Excluded",
                                       participant_id = "Completed the study"),
                            side_box = c("exclusion"),
                            cex = 0.9,
                            pch = "\u25CF")

# 后续拼接导出代码同原示例

方法3：使用gridGraphics更精准转换base绘图为grob #

gridGraphics包提供更完整的base绘图转grob能力，替代ggplotify::as.grob()：

library(gridGraphics)

# 转换base绘图为grob
grob1 <- grid.grabExpr(plot(flowchart_a))
grob2 <- grid.grabExpr(plot(flowchart_b))

# 后续拼接导出代码同原示例

内容的提问来源于stack exchange，提问作者vlangen