如何根据metadata的Patient列合并dataframe样本并计算行均值

模拟数据加载 #

先还原你提供的模拟数据：

# 基因表达数据df1
df1 <- structure(list(Sample1 = c(102, 34, 77, 13, 10), 
                      Sample2 = c(1000, 23, 199, 0, 1), 
                      Sample3 = c(293, 234, 891, 230, 2002), 
                      Sample4 = c(672, 756, 431, 293, 0), 
                      Sample5 = c(28, 2910, 0, 120, 129)), 
                 class = "data.frame", row.names = c("GeneA", "GeneB", "GeneC", "GeneD", "GeneE"))

# 元数据df2
df2 <- structure(list(Age = c(54, 70, 49, 23, 90), 
                      Patient = c(10, 10, 42, 108, 20)), 
                 class = "data.frame", row.names = c("Sample1", "Sample2", "Sample3", "Sample4", "Sample5"))

方法一：Base R 实现（无需额外包） #

通过转置、分组聚合、再转置的方式完成需求：

# 转置df1，将样本转为行
df1_transposed <- t(df1)
# 合并Patient信息
combined_data <- cbind(df1_transposed, Patient = df2$Patient)
# 按Patient分组计算基因表达均值
patient_means <- aggregate(. ~ Patient, data = combined_data, FUN = mean)
# 调整结构，恢复基因作为行名
result_base <- t(patient_means[, -1])
rownames(result_base) <- rownames(df1)
colnames(result_base) <- paste0("Patient_", patient_means$Patient)

# 输出结果
result_base

方法二：tidyverse 实现（代码更简洁） #

借助dplyr和tidyr的管道语法，可读性更强，适合处理大型数据集：

library(tidyverse)

result_tidy <- df1 %>%
  # 将基因名转为列，方便后续处理
  rownames_to_column("Gene") %>%
  # 转为长格式数据
  pivot_longer(cols = -Gene, names_to = "Sample", values_to = "Expression") %>%
  # 合并元数据中的Patient信息
  left_join(df2 %>% rownames_to_column("Sample"), by = "Sample") %>%
  # 按基因和Patient分组，计算均值
  group_by(Gene, Patient) %>%
  summarise(Mean_Expression = mean(Expression), .groups = "drop") %>%
  # 转回宽格式，恢复基因作为行名
  pivot_wider(names_from = Patient, values_from = Mean_Expression, names_prefix = "Patient_") %>%
  column_to_rownames("Gene")

# 输出结果
result_tidy

两种方法输出结果一致，最终得到的dataframe中，每一列对应一个Patient，每行是对应基因在该Patient所有样本中的表达均值。

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