如何将列更改为基于分隔符的值列表

您没有固定的分隔符,但使用 strpslit 我们可以分两部分 ID , ; , | ),则对于每个元素,仅保留以“ENSG”开头的值,并删除其他值。

sapply(strsplit(df$ID, ",|\\||;"), 
          function(x) toString(grep("^ENSG", x, value = TRUE)))


#[1] "ENSG00000143702, ENSG00000143702"                 
#[2] "ENSG00000205740"                                  
#[3] "ENSG00000067064, ENSG00000067064, ENSG00000067064"
#[4] "ENSG00000177030"      

tidyverse

library(tidyverse)
df %>%
    mutate(Genes = map_chr(str_split(ID, ";"), ~toString(map(str_split(.x, "\\|"), 2)))) %>%
    select(-ID)
#  SampleID Chrom     Start       End
#1   HSB275  chr1 243216377 243219494
#2   HSB274 chr10    952208    979839
#3   HSB272 chr10   1046378   1047984
#4   HSB481 chr11    654157    655184
#                                              Genes
#1                  ENSG00000143702, ENSG00000143702
#2                                   ENSG00000205740
#3 ENSG00000067064, ENSG00000067064, ENSG00000067064
#4                                   ENSG00000177030

样本数据

df <- read.table(text =
    "SampleID    Chrom   Start       End         ID
HSB275      chr1    243216377   243219494   ENST00000366542|ENSG00000143702|protein_coding|protein_coding,chr1,243216377,243219494;ENST00000366543|ENSG00000143702|protein_coding|protein_coding,chr1,243216377,243219494
HSB274      chr10   952208      979839      ENST00000381466|ENSG00000205740|antisense|processed_transcript,chr10,971146,979839
HSB272      chr10   1046378     1047984     ENST00000381344|ENSG00000067064|protein_coding|protein_coding,chr10,1046378,1047984;ENST00000491735|ENSG00000067064|processed_transcript|protein_coding,chr10,1046378,1047984;ENST00000427898|ENSG00000067064|protein_coding|protein_coding,chr10,1046378,1047984
HSB481      chr11   654157      655184      ENST00000527170|ENSG00000177030|nonsense_mediated_decay|protein_coding,chr11,654157,655184", header = T)

library(dplyr)
library(stringr) #str_extract_all
df %>% group_by(SampleID) %>% #Use rowwise() if you do not like group_by
       mutate(Genes = paste(str_extract_all(ID, 'ENSG\\d+',simplify = T),collapse = ',')) %>% 
       select(-ID)

# A tibble: 4 x 5
# Groups:   SampleID [4]
SampleID Chrom     Start       End Genes                                          
<fct>    <fct>     <int>     <int> <chr>                                          
1 HSB275   chr1  243216377 243219494 ENSG00000143702,ENSG00000143702                
2 HSB274   chr10    952208    979839 ENSG00000205740                                
3 HSB272   chr10   1046378   1047984 ENSG00000067064,ENSG00000067064,ENSG00000067064
4 HSB481   chr11    654157    655184 ENSG00000177030  

我的尝试:

genes %>% 
  mutate_at(vars(ID), funs(str_extract_all(., "ENSG[:digit:]*") %>% 
                                     str_replace_all("c|\"|\\(|\\)", "")))

    # A tibble: 4 x 5
  SampleID Chrom     Start       End ID                                               
  <chr>    <chr>     <dbl>     <dbl> <chr>                                            
1 HSB275   chr1  243216377 243219494 ENSG00000143702, ENSG00000143702                 
2 HSB274   chr10    952208    979839 ENSG00000205740                                  
3 HSB272   chr10   1046378   1047984 ENSG00000067064, ENSG00000067064, ENSG00000067064
4 HSB481   chr11    654157    655184 ENSG00000177030   

这将查找与匹配的任何模式 ENSG<any length of numeric characters> ,然后将列表强制为相关字符串的向量,并整理所有不需要的字符。

虽然我个人的精神是整洁的数据,但我会将每个“ID”放在一个单独的列中,复制相关的SampleID/Chrom/Start/End数据。

数据

df <- read.table(text =
    "SampleID    Chrom   Start       End         ID
HSB275      chr1    243216377   243219494   ENST00000366542|ENSG00000143702|protein_coding|protein_coding,chr1,243216377,243219494;ENST00000366543|ENSG00000143702|protein_coding|protein_coding,chr1,243216377,243219494
HSB274      chr10   952208      979839      ENST00000381466|ENSG00000205740|antisense|processed_transcript,chr10,971146,979839
HSB272      chr10   1046378     1047984     ENST00000381344|ENSG00000067064|protein_coding|protein_coding,chr10,1046378,1047984;ENST00000491735|ENSG00000067064|processed_transcript|protein_coding,chr10,1046378,1047984;ENST00000427898|ENSG00000067064|protein_coding|protein_coding,chr10,1046378,1047984
HSB481      chr11   654157      655184      ENST00000527170|ENSG00000177030|nonsense_mediated_decay|protein_coding,chr11,654157,655184", header = T)

我的解决方案

我定义了两个函数,这两个函数将来可能会有用,并且可以解决此任务:

第一,, extract_matches ,提取中每个元素的所有匹配项 str.vec gregexpr 它只返回匹配的位置信息。

第二,, extract_matches_aggregating sep= . 这取决于 提取匹配项 .

您可以使用这两个函数提取所有ENSG ID,并通过“,”链接它们。

extract_matches <- function(pattern, str.vec) {
  Map(function(m, s) substring(s, m, m + attr(m, "match.length") - 1), gregexpr(pattern, str.vec), str.vec)
}

extract_matches_aggregating <- function(pattern, str.vec, sep = "; ") {
  sapply(extract_matches(pattern, str.vec), function(res_vec) {
           paste(res_vec, collapse = sep)})
}

df$ID <- extract_matches_aggregating(pattern = "ENSG\\d+", str.vec = df$ID, sep = ", ")

df 那么:

  SampleID Chrom     Start       End
1   HSB275  chr1 243216377 243219494
2   HSB274 chr10    952208    979839
3   HSB272 chr10   1046378   1047984
4   HSB481 chr11    654157    655184
                                                 ID
1                  ENSG00000143702, ENSG00000143702
2                                   ENSG00000205740
3 ENSG00000067064, ENSG00000067064, ENSG00000067064
4                                   ENSG00000177030

在大型表上,此解决方案比使用 strsplit 和 sapply 和 lapply