如何编写循环批量读取非连续ID的神经元.dat文件并生成指定数据框

现有76个神经元，每个对应两个.dat格式文件（文件名前缀为非连续cell ID，比如1_perDend_counts.dat和1_perDend.dat），总计152个文件。需要实现批量读取所有文件，合并成包含以下列的单一数据框：cell_ID, tuft avg, basal avg, oblique avg, prim avg, tuft order avg, basal order avg, tuft count, basal count, oblique count，且各平均值与对应cell ID绑定。

你提供的单个神经元处理代码如下：

count1 <- read.table("1_perDend_counts.dat", header = T)
morph1 <- read.table("1_perDend.dat", header = T)

morph1$order <- as.numeric(morph1$order)

avg_data <- morph1 %>%
  group_by(cell_id) %>%
  mutate(tuft_avg = mean(length[dend_type == "tuft"]),
         basal_avg = mean(length[dend_type == "basal"]),
         oblique_avg = mean(length[dend_type == "oblique"]),
         prim_avg = mean(length[dend_type == "primary"]),
         tuft_order_avg = mean(order[dend_type == "tuft"]),
         basal_order_avg = mean(order[dend_type == "basal"])) %>%
  slice(1) %>%
  select(-order, -dend_type, -length)

count1 <- spread(count1, dend_type, count)
colnames(count1) <- c("cell_id", "tuft_count", "basal_count", "oblique_count")

all <- merge(avg_data, count1, by = "cell_id", all = TRUE)

以下是扩展到所有神经元的批量处理方案：

实现步骤 #

1. 自动提取所有cell ID #

无需手动列出所有ID，从文件名中自动提取：

# 获取目录中所有目标文件
count_files <- list.files(pattern = "_perDend_counts.dat$")
morph_files <- list.files(pattern = "_perDend.dat$")

# 提取唯一的cell ID
cell_ids <- unique(sub("_perDend.*\\.dat$", "", c(count_files, morph_files)))

2. 封装单个cell的处理函数 #

把单个神经元的处理逻辑打包成函数，方便循环调用：

library(dplyr)
library(tidyr)

process_single_cell <- function(cell_id) {
  # 构建文件路径
  count_path <- paste0(cell_id, "_perDend_counts.dat")
  morph_path <- paste0(cell_id, "_perDend.dat")
  
  # 读取数据
  count_df <- read.table(count_path, header = TRUE)
  morph_df <- read.table(morph_path, header = TRUE)
  
  # 处理形态数据，计算各类平均值
  morph_df$order <- as.numeric(morph_df$order)
  
  avg_df <- morph_df %>%
    group_by(cell_id) %>%
    mutate(
      tuft_avg = mean(length[dend_type == "tuft"], na.rm = TRUE),
      basal_avg = mean(length[dend_type == "basal"], na.rm = TRUE),
      oblique_avg = mean(length[dend_type == "oblique"], na.rm = TRUE),
      prim_avg = mean(length[dend_type == "primary"], na.rm = TRUE),
      tuft_order_avg = mean(order[dend_type == "tuft"], na.rm = TRUE),
      basal_order_avg = mean(order[dend_type == "basal"], na.rm = TRUE)
    ) %>%
    slice(1) %>%
    select(-order, -dend_type, -length)
  
  # 处理计数数据，调整列结构
  count_df <- count_df %>%
    spread(dend_type, count) %>%
    rename(
      tuft_count = tuft,
      basal_count = basal,
      oblique_count = oblique
    )
  
  # 合并两类数据
  merged_df <- merge(avg_df, count_df, by = "cell_id", all = TRUE)
  return(merged_df)
}

3. 批量处理并合并结果 #

用lapply遍历所有cell ID，再将结果合并为一个数据框：

# 批量处理所有神经元
final_data <- lapply(cell_ids, process_single_cell) %>%
  bind_rows()

# 查看结果示例
head(final_data)

补充说明 #

• 确保已安装dplyr和tidyr包，未安装的话先运行：install.packages(c("dplyr", "tidyr"))
• 添加na.rm = TRUE是为了避免因部分神经元缺失某类树突导致平均值计算为NA
• 如果文件不在当前工作目录，需要在list.files中指定path参数，比如list.files(path = "your/data/dir", pattern = "...")

内容的提问来源于stack exchange，提问作者Minda