R语言：如何移除dataframe中gs_name列匹配多子串的行？

我有一个名为all.gene.sets的数据框，其中gs_name列的字符串可能匹配redundant.gs向量中的任意子串。我需要移除所有存在此类匹配的行，但当前代码仅能移除匹配第一个子串的行，无法处理后续子串。

当前代码：

redundant.gs <- c("ANDERSON_BLOOD_CN54GP140_ADJUVANTED", "BUCASAS_PBMC_FLUARIX_FLUVIRIN")
gene.sets <- all.gene.sets[!(all.gene.sets$gs_name %like% redundant.gs),]

测试数据：

all.gene.sets <- tibble::tribble(
  ~gs_cat, ~gs_subcat,    ~gs_name,
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_DN",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_DN",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_DN",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_DN",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_DN",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_UP",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_UP",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_UP",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_UP",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_UP",
  "C7",    "VAX",         "ANDERSON_BLOOD_CN54GP140_ADJUVANTED_WITH_GLA_AF_AGE_18_45YO_1DY_UP",
  "C7",    "VAX",         "BUCASAS_PBMC_FLUARIX_FLUVIRIN_CAUCASIAN_MALE_AGE_18_40YO_HIGH_RESPONDERS_1DY_3DY_POSITIVE_PREDICTIVE_OF_TITER",
  "C7",    "VAX",         "BUCASAS_PBMC_FLUARIX_FLUVIRIN_CAUCASIAN_MALE_AGE_18_40YO_HIGH_RESPONDERS_1DY_3DY_POSITIVE_PREDICTIVE_OF_TITER",
  "C7",    "VAX",         "BUCASAS_PBMC_FLUARIX_FLUVIRIN_CAUCASIAN_MALE_AGE_18_40YO_HIGH_RESPONDERS_1DY_3DY_POSITIVE_PREDICTIVE_OF_TITER",
  "C7",    "VAX",         "BUCASAS_PBMC_FLUARIX_FLUVIRIN_CAUCASIAN_MALE_AGE_18_40YO_HIGH_RESPONDERS_1DY_3DY_POSITIVE_PREDICTIVE_OF_TITER",
  "C7",    "VAX",         "BUCASAS_PBMC_FLUARIX_FLUVIRIN_CAUCASIAN_MALE_AGE_18_40YO_HIGH_RESPONDERS_1DY_3DY_POSITIVE_PREDICTIVE_OF_TITER",
  "C7",    "IMMUNESIGDB", "GSE21063_3H_VS_16H_ANTI_IGM_STIM_NFATC1_KOBCELL_DN",
  "C7",    "IMMUNESIGDB", "GSE21063_3H_VS_16H_ANTI_IGM_STIM_NFATC1_KOBCELL_DN",
  "C7",    "IMMUNESIGDB", "GSE21063_3H_VS_16H_ANTI_IGM_STIM_NFATC1_KOBCELL_DN",
  "C7",    "IMMUNESIGDB", "GSE21063_3H_VS_16H_ANTI_IGM_STIM_NFATC1_KOBCELL_DN",
  "C7",    "IMMUNESIGDB", "GSE21063_3H_VS_16H_ANTI_IGM_STIM_NFATC1_KOBCELL_DN",
)

原来的%like%操作符（常见于data.table或dplyr包）当右侧传入向量时，只会匹配第一个元素，因此无法处理多个子串。可以通过以下几种方法解决：

方法1：基础R + 正则表达式合并模式 #

将redundant.gs中的子串用|连接成一个正则表达式模式，再用grepl检测匹配：

redundant.pattern <- paste(redundant.gs, collapse = "|")
gene.sets <- all.gene.sets[!grepl(redundant.pattern, all.gene.sets$gs_name), ]

方法2：dplyr + stringr 管道风格 #

如果你习惯用dplyr的管道语法，可以结合str_detect和filter实现：

library(dplyr)
library(stringr)

gene.sets <- all.gene.sets %>%
  filter(!str_detect(gs_name, paste(redundant.gs, collapse = "|")))

方法3：data.table 多模式处理 #

如果使用data.table，可直接传入合并后的正则表达式，或用sapply结合any()检测每个子串：

library(data.table)
setDT(all.gene.sets)

# 方式A：合并正则表达式
gene.sets <- all.gene.sets[!gs_name %like% paste(redundant.gs, collapse = "|")]

# 方式B：逐行检测所有子串
gene.sets <- all.gene.sets[!sapply(gs_name, function(x) any(x %like% redundant.gs))]

以上方法都会移除gs_name列包含redundant.gs中任意子串的行，最终保留5行IMMUNESIGDB相关数据。

内容的提问来源于stack exchange，提问作者Anon