if, ifelse, or apply in R loop to split and format datasets

Question

I am reformatting ~150 files in R. I have been doing this with code provided below, but I now realize that I need to split each of these files into 2 files based on a column variable, and then do the reformatting. I am stuck with the splitting! Here's the process:

I have a folder of files and I am reading file names into R like so:

setwd("/home/intersected_beds")
path <-("/home/intersected_beds")
data <- dir(path)

All of the files look something like this, with varying number of rows but columns matching exactly:

   chr     start       end dir subfamily  family
1 chr1  87144764  87150794   C      L1HS LINE/L1
2 chr2 173179999 173186025   +      L1HS LINE/L1
3 chr2 181698389 181704416   C      L1HS LINE/L1
4 chr3 108468248 108474272   +      L1HS LINE/L1
5 chr3 132664851 132670878   C      L1HS LINE/L1
6 chr4  53682624  53688653   +      L1HS LINE/L1

I need to reformat the data so that it looks like this:

    chr    start      end                  unique
1 chr11 20363925 20370314 chr11_20363925-20370314
2 chr13 46788764 46795064 chr13_46788764-46795064
etc

Which I have been doing like this:

for(i in data){
  t <- read.table(i, header = FALSE, stringsAsFactors = FALSE) # load file
  colnames(t) <- c("chr", "start", "end", "dir", "subfamily", "family")
  fmt <- paste(t$chr,"_", t$start,"-", t$end)
  fmt <- gsub(" ","", fmt)
  fmt <- as.data.frame(fmt)
  new.t <- cbind(t, fmt)
  colnames(new.t) <- c("chr", "start", "end", "dir", "subfamily", "family", "unique")
  new.t.bed <- new.t %>% select(chr, start, end, unique)
  write.table(new.t.bed, paste0(tools::file_path_sans_ext(i), "_2bit", ".bed"), col.names=FALSE, row.names=FALSE, quote = FALSE)
}

However, it turns out that I need to split my data based on the "dir" column in the input data, and then do this reformatting. The dir column consists of + or c values. I think I need an ifelse statement because I ran into "the condition has length > 1 and only the first element will be used" errors with a regular if statement. I've attempted to split and format them like so:

for(i in data){
  t <- read.table(i, header = FALSE, stringsAsFactors = FALSE)
  colnames(t) <- c("chr", "start", "end", "dir", "subfamily", "family")
  if(t$dir =="+"){
    fmt <- paste(t$chr,"_", t$start,"-", t$end)
    fmt <- gsub(" ","", fmt)
    fmt <- as.data.frame(fmt)
    new.t <- cbind(t, fmt)
    colnames(new.t) <- c("chr", "start", "end", "dir", "subfamily", "family", "unique")
    new.t.bed <- new.t %>% select(chr, start, end, unique)
    write.table(new.t.bed, paste0(tools::file_path_sans_ext(i), "_2bit_plus", ".bed"), col.names=FALSE, row.names=FALSE, quote = FALSE)
  } else {
    fmt <- paste(t$chr,"_", t$start,"-", t$end)
    fmt <- gsub(" ","", fmt)
    fmt <- as.data.frame(fmt)
    new.t <- cbind(t, fmt)
    colnames(new.t) <- c("chr", "start", "end", "dir", "subfamily", "family", "unique")
    new.t.bed <- new.t %>% select(chr, start, end, unique)
    write.table(new.t.bed, paste0(tools::file_path_sans_ext(i), "_2bit_c", ".bed"), col.names=FALSE, row.names=FALSE, quote = FALSE)
  }
}

but I am still getting a warning "the condition has length > 1 and only the first element will be used" and the output does not have the data split by + and c. I thought about trying to make plussplit and csplit functions to use in an ifelse statement, but I got the error "Error in dir[[t]] : object of type 'closure' is not subsettable"

for(i in data){
  t <- read.table(i, header = FALSE, stringsAsFactors = FALSE)
  colnames(t) <- c("chr", "start", "end", "dir", "subfamily", "family")
  ifelse(dir[[t]] =="+", plussplit, csplit)
}
  
plussplit <- function(t){
  fmt <- paste(t$chr,"_", t$start,"-", t$end)
  fmt <- gsub(" ","", fmt)
  fmt <- as.data.frame(fmt)
  new.t <- cbind(t, fmt)
  colnames(new.t) <- c("chr", "start", "end", "dir", "subfamily", "family", "unique")
  new.t.bed <- new.t %>% select(chr, start, end, unique)
  write.table(new.t.bed, paste0(tools::file_path_sans_ext(i), "_2bit_plus", ".bed"), col.names=FALSE, row.names=FALSE, quote = FALSE)
}

csplit <- function(t){
  fmt <- paste(t$chr,"_", t$start,"-", t$end)
  fmt <- gsub(" ","", fmt)
  fmt <- as.data.frame(fmt)
  new.t <- cbind(t, fmt)
  colnames(new.t) <- c("chr", "start", "end", "dir", "subfamily", "family", "unique")
  new.t.bed <- new.t %>% select(chr, start, end, unique)
  write.table(new.t.bed, paste0(tools::file_path_sans_ext(i), "_2bit_c", ".bed"), col.names=FALSE, row.names=FALSE, quote = FALSE)
}

I've looked through the posts on these topics and see the apply functions being suggested, but I can't get anything to work. I imagine there is something fundamental I am missing about this process, and I bet you royal beings of stackoverflow have the answers! Thanks!

Answer 1

Not sure if I got it right. You could use

library(dplyr)
library(purrr)

for(i in data) {
  t <- read.table(i, header = FALSE, stringsAsFactors = FALSE)
  
  colnames(t) <- c("chr", "start", "end", "dir", "subfamily", "family")
  
  t %>% 
    mutate(unique = gsub(" ", "", paste(chr, start, end, sep = "_"))) %>% 
    select(dir, chr, start, end, unique) %>% 
    split(f = .$dir) %>% 
    walk(~ifelse(
      all(.x$dir == "+"), 
      .x %>% 
        select(-dir) %>% 
        write.table(paste0(tools::file_path_sans_ext(i), "_2bit_plus", ".bed"), 
                    col.names = FALSE, 
                    row.names = FALSE, 
                    quote = FALSE), 
      .x %>% 
        select(-dir) %>% 
        write.table(paste0(tools::file_path_sans_ext(i), "_2bitc", ".bed"), 
                    col.names = FALSE, 
                    row.names = FALSE, 
                    quote = FALSE)))
}

• The mutate-function is basically your creating of fmt, just in one step.
• split splits the data.frame by the values of a given variable, in this case dir. Since there are two unique values of dir, you get a list of two data.frames: one for + and another one for c.
• walk from package purrr now applies an ifelse-function on those data.frame: if all values of dir are + then a file name containing _2bit_plus.bed is created else _2bit_c.bed.
• Be careful to run this code. You might overwrite existing files that you wanted to keep.