R语言中如何移除匹配多个模式的多行数据？解决grepl仅识别首个模式的问题

Hey there! The issue you're seeing is because grepl() doesn't accept a vector of patterns directly—when you pass more than one, it only uses the first one (which is exactly what the warning is telling you). Here's how to fix this:

The Quick Fix for Base R #

Combine all your patterns into a single string separated by | (which means "OR" in regex). You can do this using paste() with collapse = "|":

# Define your patterns once for clarity
phylum_patterns <- c("Chloroflexota", "Desulfobacterota_D", "Gemmatimonadota")

# Combine into one regex pattern that matches any of the entries
combined_pattern <- paste(phylum_patterns, collapse = "|")

# Filter the data frame to exclude rows matching any of the patterns
custom_BGCs_filtered <- custom_BGCs[!grepl(combined_pattern, custom_BGCs$Phylum), ]

If you prefer a concise one-liner:

custom_BGCs_filtered <- custom_BGCs[!grepl(paste(c("Chloroflexota","Desulfobacterota_D", "Gemmatimonadota"), collapse = "|"), custom_BGCs$Phylum), ]

Bonus: Tidyverse Approach (dplyr + stringr) #

If you use the tidyverse ecosystem, filter() paired with str_detect() makes this even more readable:

library(dplyr)
library(stringr)

custom_BGCs_filtered <- custom_BGCs %>%
  filter(!str_detect(Phylum, paste(phylum_patterns, collapse = "|")))

If you want to treat your phylum names as literal strings (no regex interpretation, just in case they contain special characters), add fixed = TRUE:

# Base R version with literal matching
custom_BGCs_filtered <- custom_BGCs[!grepl(combined_pattern, custom_BGCs$Phylum, fixed = TRUE), ]

# Tidyverse version with literal matching
custom_BGCs_filtered <- custom_BGCs %>%
  filter(!str_detect(Phylum, fixed(phylum_patterns)))

Why This Works #

By combining patterns with |, you're telling the function to look for any of the specified phyla in the Phylum column. The ! operator inverts the match, so we keep only the rows that don't match any of your target patterns.

内容的提问来源于stack exchange，提问作者Valentin