基于probeID匹配coMeth_ls的cg*项筛选数据框返回空值求助

问题场景 #

想要对数据框distal.results.stratified.model.1a做子集筛选，保留probeID列值匹配列表coMeth_ls中任意cg*格式项的行，但执行以下代码后返回空数据框：

distal.dmr.cpgs.1a <- distal.results.stratified.model.1a[
  sapply(
    strsplit(distal.results.stratified.model.1a$probeID, ' '), 
    function(x) any(x %in% coMeth_ls)), ]

数据详情 #

> dput(distal.results.stratified.model.1a[1:3,1:3])
structure(list(regionID = c("chr1:1355332-1355333", "chr1:15934516-15934517", 
"chr1:15934516-15934517"), probeID = c("cg03324138", "cg01522970", 
"cg01522970"), target = c("ENSG00000162576", "ENSG00000065526", 
"ENSG00000065526")), row.names = c(NA, 3L), class = "data.frame")

> dput(coMeth_ls[1:10])
list(`chr1:2250410-2250796` = c("cg03324138", "cg02237208", "cg00567930"
), `chr1:2886792-2886868` = c("cg01977798", "cg00050294", "cg04080530"
), `chr1:8271997-8272277` = c("cg02847588", "cg00590817", "cg06780032"
), `chr1:10895950-10896004` = c("cg02971581", "cg09510180", "cg00944431"
), `chr1:12600225-12600529` = c("cg07586026", "cg07242540", "cg00003578"
), `chr1:16553267-16553456` = c("cg03969515", "cg03522107", "cg02987928"
), `chr1:19665131-19665240` = c("cg09356672", "cg05348366", "cg04928049"
), `chr1:23003462-23003569` = c("cg01795776", "cg09062638", "cg02623028"
), `chr1:28573736-28573940` = c("cg05045517", "cg03226554", "cg07148744", 
"cg07913096"), `chr1:29460817-29461070` = c("cg06155303", "cg09264065", 
"cg08597839", "cg00107241"))

错误原因 #

1. 多余的strsplit操作：你的probeID列是单个cg*字符串，没有空格分隔，拆分后得到的是单元素向量，这步完全没必要，属于冗余操作。
2. 核心匹配错误：coMeth_ls是嵌套列表，每个元素是对应区域的cg*向量。你直接用x %in% coMeth_ls，是拿单个cg值和整个列表元素（向量）做匹配，永远不会返回TRUE，自然筛选不到任何行。

解决办法 #

先把嵌套列表里的所有cg*值提取成一个扁平的向量，再用这个向量做匹配筛选。

基础R实现

# 提取所有cg*值到扁平向量
all_cg_values <- unlist(coMeth_ls)
# 直接筛选，无需strsplit
distal.dmr.cpgs.1a <- distal.results.stratified.model.1a[
  distal.results.stratified.model.1a$probeID %in% all_cg_values, 
]

dplyr实现（更直观）

library(dplyr)

all_cg_values <- unlist(coMeth_ls)
distal.dmr.cpgs.1a <- distal.results.stratified.model.1a %>%
  filter(probeID %in% all_cg_values)

执行后就能得到包含probeID为cg03324138的行，符合预期。

内容的提问来源于stack exchange，提问作者Anon