CODEWITHSUNDEEP
r
dsperley
dsperley

Reputation: 43

Splitting based on delimiter, and removing duplicates

I have dataframe (Annot_Subset) that looks like this:

IlmnID   UCSC_RefGene_Group
cg00050873         Body;TSS1500
cg00212031               TSS200
cg00213748                     
cg00214611        1stExon;5'UTR
cg00455876                     
cg01707559 TSS200;TSS200;TSS200
cg02004872        1stExon;5'UTR

What I would like to do is clean up the last column by doing the following:

  1. Replace entries with different strings (ex. Body; TSS1500) with the string "Multiple_locations"

  2. Condense entries with same string repeated (ex. TSS200; TSS200) with that string appearing once

  3. Adding the string "Intergenic" to entries that contain nothing.

To give an example:

IlmID UCSC_RefGene_Group
cg00050873 Multiple_locations
cg00212031             TSS200
cg00213748         Intergenic
cg00214611 Multiple_locations
cg00455876         Intergenic
cg01707559             TSS200
cg02004872 Multiple_locations

I have written a function that will do this; but was wondering if there was a more elegant and efficient way to approach the problem. Especially since my dataframe has 485 000 rows.

This is what I have come up with:

Gene_Group_Split<-strsplit(Annot_Subset$UCSC_RefGene_Group,";")
Clean.Gene.Group<-function(x) {
  Gene_Group_Cleaned<-vector(mode="character",length=length(x))
  for (i in 1:length(x)) { 
    if (length(x[[i]])>=1) {
      unique_set<-unique(x[[i]])
      if (length(unique_set)==1) {
        Gene_Group_Cleaned[i]<-unique_set
      } else {
        Gene_Group_Cleaned[i]<-"Multiple_locations"
      }
    } else {
      Gene_Group_Cleaned[i]<-"Intergenic"
    }
  }  
return(Gene_Group_Cleaned)
}

Gene_Group_2<-Clean.Gene.Group(Gene_Group_Split)

Upvotes: 1

Views: 109

Answers (2)

bramtayl
bramtayl

Reputation: 4024

library(dplyr)
library(stringi)
library(tidyr)

df = read.table(text="
IlmnID UCSC_RefGene_Group
                 cg00050873 Body;TSS1500
                 cg00212031 TSS200
                 cg00213748                    
                 cg00214611 1stExon;5'UTR
                 cg00455876                    
                 cg01707559 TSS200;TSS200;TSS200
                 cg02004872 1stExon;5'UTR", fill=TRUE, header=TRUE, stringsAsFactors=FALSE)

classify = function(string_vector)
  if (length(string_vector) > 1) "Multiple_locations" else
    if (string_vector == "") "Intergenic" else
      string_vector

df %>%
  mutate(UCSC_RefGene_Group = 
           UCSC_RefGene_Group %>% 
           stri_split_fixed(";")) %>%
  unnest(UCSC_RefGene_Group) %>%
  distinct %>%
  group_by(IlmnID) %>%
  summarize(class = classify(UCSC_RefGene_Group))

Upvotes: 0

zx8754
zx8754

Reputation: 56149

Try this example:

df <- read.table(text="
IlmnID UCSC_RefGene_Group
                 cg00050873 Body;TSS1500
                 cg00212031 TSS200
                 cg00213748                    
                 cg00214611 1stExon;5'UTR
                 cg00455876                    
                 cg01707559 TSS200;TSS200;TSS200
                 cg02004872 1stExon;5'UTR", fill=TRUE, header=TRUE, stringsAsFactors=FALSE)

df$Type <-
  unlist(
    lapply(df$UCSC_RefGene_Group, function(i){
      x <- unique(unlist(strsplit(i,split = ";")))
      ifelse(length(x)>1,"Multiple_locations",
             ifelse(length(x)==0,"Intergenic",x))
    })
  )

#result
df
#       IlmnID   UCSC_RefGene_Group               Type
# 1 cg00050873         Body;TSS1500 Multiple_locations
# 2 cg00212031               TSS200             TSS200
# 3 cg00213748                              Intergenic
# 4 cg00214611        1stExon;5'UTR Multiple_locations
# 5 cg00455876                              Intergenic
# 6 cg01707559 TSS200;TSS200;TSS200             TSS200
# 7 cg02004872        1stExon;5'UTR Multiple_locations

Upvotes: 2

Related Questions