ggplot2添加文本标签时数据点错位问题排查与解决

在使用ggplot2绘制散点图并添加文本标签时，出现标签与对应数据点错位的问题——例如Kynurenine的log2FC值为3.6601（正值），但其标签却显示在负值区域。简化后的复现代码如下：

library(ggplot2)
library(ggrepel)
library(dplyr)

df = data.frame(structure(list(FC = c(4.712, 4.8264, 0.078026, 0.098754, 11.612, 
0.21181, 6.1715, 0.15113, 0.19936, 0.23572, 0.23564, 6.9851, 
5.2465, 12.641, 4.4164, 0.24834, 7.4598, 5.1384, 0.18861, 0.24986
), log2FC = c(2.2363, 2.271, -3.6799, -3.34, 3.5375, -2.2391, 
2.6256, -2.7261, -2.3266, -2.0849, -2.0853, 2.8043, 2.3914, 3.6601, 
2.1429, -2.0096, 2.8991, 2.3613, -2.4065, -2.0008), Identifier = 1:20), row.names = c("3-Indoleacetic acid", 
"4-Hydroxyphenylpyruvic acid", "Adenine", "Adenosine", "alpha-Ketoglutaric acid", 
"C10", "C18:1OH", "Deoxyadenosine", "Deoxycytidine", "Deoxyguanosine", 
"Histamine", "Homocitrulline", "Indolelactic acid", "Kynurenine", 
"Leucine", "LysoPC a C20:4", "N-Acetylputrescine", "Phenylalanine", 
"Phenylethylamine", "Spermidine"), class = "data.frame"))


show.names = c("Kynurenine", "alpha-Ketoglutaric acid", "N-Acetylputrescine", "Homocitrulline", "C18:1OH", "Deoxycytidine", "Phenylethylamine", "Deoxyadenosine", "Adenosine", "Adenine")

p1 = ggplot(df, aes(x = Identifier, y = log2FC)) +
     geom_point(aes(color = log2FC >= 0)) +
     scale_color_manual(name = "Group", values = c("red", "blue"), labels = c("Down", "Up")) +
     geom_text_repel(data = df[rownames(df) %in% show.names,], aes(label = show.names)) +
     geom_hline(yintercept = 0, linetype = "solid") +
     labs(x = "Compounds", y = "Log2 Fold Change") + theme_minimal()
p1

原因分析 #

错位的核心问题在于geom_text_repel中的label参数设置：

• df[rownames(df) %in% show.names,]筛选出的数据集，其行顺序是按照原数据框的行序排列（比如筛选后的第一行是Adenine，第二行是Adenosine，而非show.names的顺序）
• 但直接将show.names作为标签向量赋值，导致标签被按顺序硬塞给筛选后的每行数据，完全不匹配对应行的实际化合物名称，最终出现标签错位。

修复方案 #

方案1：直接使用筛选后数据集的行名作为标签 #

将label参数改为筛选后数据框的行名，确保标签与对应数据行匹配：

p1 = ggplot(df, aes(x = Identifier, y = log2FC)) +
     geom_point(aes(color = log2FC >= 0)) +
     scale_color_manual(name = "Group", values = c("red", "blue"), labels = c("Down", "Up")) +
     # 用筛选后数据的行名作为标签
     geom_text_repel(data = df[rownames(df) %in% show.names,], 
                    aes(label = rownames(df[rownames(df) %in% show.names,]))) +
     geom_hline(yintercept = 0, linetype = "solid") +
     labs(x = "Compounds", y = "Log2 Fold Change") + theme_minimal()
p1

方案2：先创建筛选后的数据集，再调用其行名 #

更清晰的写法是先将筛选后的数据存为单独变量，再引用该变量的行名，避免重复代码：

# 先筛选需要加标签的数据
label_data <- df[rownames(df) %in% show.names,]

p1 = ggplot(df, aes(x = Identifier, y = log2FC)) +
     geom_point(aes(color = log2FC >= 0)) +
     scale_color_manual(name = "Group", values = c("red", "blue"), labels = c("Down", "Up")) +
     geom_text_repel(data = label_data, aes(label = rownames(label_data))) +
     geom_hline(yintercept = 0, linetype = "solid") +
     labs(x = "Compounds", y = "Log2 Fold Change") + theme_minimal()
p1

方案3：给原数据框添加标记列，按需显示标签 #

如果需要多次复用标记逻辑，可以给原数据框加一列标记哪些化合物需要显示标签，再通过subset筛选：

# 添加标记列
df$need_label <- rownames(df) %in% show.names

p1 = ggplot(df, aes(x = Identifier, y = log2FC)) +
     geom_point(aes(color = log2FC >= 0)) +
     scale_color_manual(name = "Group", values = c("red", "blue"), labels = c("Down", "Up")) +
     geom_text_repel(data = subset(df, need_label), aes(label = rownames(df))) +
     geom_hline(yintercept = 0, linetype = "solid") +
     labs(x = "Compounds", y = "Log2 Fold Change") + theme_minimal()
p1

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