如何使用ggsankey自定义桑基图第二层节点的显示顺序？

问题描述 #

我使用R语言的ggsankey包绘制了桑基图，整体效果正常，但希望调整第二层节点的排列顺序：当前节点从上到下为SD、PR、DP、CR，需改为CR、PR、SD、DP。请问是否有自定义节点位置的方法？

原代码 #

devtools::install_github("davidsjoberg/ggsankey")
library(ggsankey)

sankey_df <- base_limpia%>% transmute(
  Treatment = Type.of.Treatment,
  Response = Respuesta,
  Transplantation = Post_therapy_trasplantation
)

sankey_df <- sankey_df %>% make_long(Treatment,
                                     Response,
                                     Transplantation
)

reagg <- sankey_df%>%
  dplyr::group_by(node)%>%  # 按节点分组统计频次
  tally()

sankey_df2 <- merge(sankey_df,
             reagg, 
             by.x = 'node', 
             by.y = 'node', 
             all.x = TRUE)
png(file = "Figure 6.png", width = 3200, height = 2000, res = 300)
plot <- ggplot(sankey_df2, aes(x = x,                        
                     next_x = next_x,                                     
                     node = node,
                     next_node = next_node,        
                     fill = factor(node),
                     label = paste0(node, " = ", n)))

plot <- plot + geom_sankey(flow.alpha = 0.5,          
                      node.color = "black",        
                      show.legend = F) + 
  geom_sankey_label(size = 3,
                    color = "black",
                    fill = "white") +
  theme_bw() +
  theme(legend.position = "none") +
  theme(axis.title = element_blank(),
        axis.text.y = element_blank(),
        axis.ticks = element_blank(),
        panel.grid = element_blank())

解决方案 #

ggsankey的节点排列顺序默认跟随对应列的因子水平或数据出现顺序，要自定义第二层（Response对应层级）的节点顺序，只需在数据预处理阶段将Response列转换为指定顺序的因子即可：

修改步骤 #

1. 在transmute生成sankey_df时，将Response列设置为因子，指定levels为目标顺序c("CR", "PR", "SD", "DP")。
2. 后续流程保持不变，make_long会沿用这个因子顺序生成桑基图节点。

修改后的完整代码 #

devtools::install_github("davidsjoberg/ggsankey")
library(ggsankey)
library(dplyr)

sankey_df <- base_limpia %>% transmute(
  Treatment = Type.of.Treatment,
  # 关键修改：将Response设为指定顺序的因子
  Response = factor(Respuesta, levels = c("CR", "PR", "SD", "DP")),
  Transplantation = Post_therapy_trasplantation
)

sankey_df <- sankey_df %>% make_long(Treatment,
                                     Response,
                                     Transplantation
)

reagg <- sankey_df %>%
  dplyr::group_by(node)%>%
  tally()

sankey_df2 <- merge(sankey_df,
             reagg, 
             by.x = 'node', 
             by.y = 'node', 
             all.x = TRUE)
png(file = "Figure 6.png", width = 3200, height = 2000, res = 300)
plot <- ggplot(sankey_df2, aes(x = x,                        
                     next_x = next_x,                                     
                     node = node,
                     next_node = next_node,        
                     fill = factor(node),
                     label = paste0(node, " = ", n)))

plot <- plot + geom_sankey(flow.alpha = 0.5,          
                      node.color = "black",        
                      show.legend = F) + 
  geom_sankey_label(size = 3,
                    color = "black",
                    fill = "white") +
  theme_bw() +
  theme(legend.position = "none") +
  theme(axis.title = element_blank(),
        axis.text.y = element_blank(),
        axis.ticks = element_blank(),
        panel.grid = element_blank())
dev.off()  # 关闭绘图设备

如果上述方法未生效，也可以在生成sankey_df2后，针对第二层（x=2）的节点单独设置因子顺序：

sankey_df2 <- sankey_df2 %>%
  mutate(node = case_when(
    x == 2 ~ factor(node, levels = c("CR", "PR", "SD", "DP")),
    TRUE ~ as.factor(node)
  ))

内容的提问来源于stack exchange，提问作者vaquera