Obtain unique no overlapping region of bed file

[rkawa]

Hello. I have a bed file that contains exons of all mouse ensembl genes. I am trying to get UNIQUE regions which have no overlapping with other ensembl genes. I thought there should be some tools for this, but I haven't been able find one that works. All the tools I found do not remove overlapping regions. Here is a simple example

chr1 10 20 gene1
chr1 15 22 gene2
chr1 20 25 gene3

the result should be
chr1 10 14 gene1
chr1 23 25 gene3

I think this is possible with algorithm using bedtools subtract. But if anyone knows there is a simpler way to do it, I highly appreciate!

Thanks,

RK

[dpryan]

An example in R:

library(GenomicRanges)
gr <- GRanges(seqnames=c(rep("1",5)), ranges=IRanges(start=c(10,15,20,25,30), end=c(20,21,30,40,35)))
dis <- disjoin(gr) #Make disjoint ranges
res <- dis[countOverlaps(dis,gr)==1] #subset by those bins overlapping only 1 original entry

In case you're wondering, the input data ("gr") looks like this:

      seqnames    ranges strand
         <Rle> <IRanges>  <Rle>
  [1]        1  [10, 20]      *
  [2]        1  [15, 21]      *
  [3]        1  [20, 30]      *
  [4]        1  [25, 40]      *
  [5]        1  [30, 35]      *

The disjoint bins ("dis") are then:

         <Rle> <IRanges>  <Rle>
  [1]        1  [10, 14]      *
  [2]        1  [15, 19]      *
  [3]        1  [20, 20]      *
  [4]        1  [21, 21]      *
  [5]        1  [22, 24]      *
  [6]        1  [25, 29]      *
  [7]        1  [30, 30]      *
  [8]        1  [31, 35]      *
  [9]        1  [36, 40]      *

So the results ("res") are:

         <Rle> <IRanges>  <Rle>
  [1]        1  [10, 14]      *
  [2]        1  [22, 24]      *
  [3]        1  [36, 40]      *

bedtools is very likely faster, but it's often nice to use R since it's convenient for debugging.

[rkawa]

Excellent. I split original bed files by chromosome names, then run the script.
After that I concatenate the resulting uniq exon bed files into one.
bedtools intersect with original bed file allows to add gene annotation.

Thanks you so much.