Bedtools Intersect -v returns same file, with only one line removed

Question

[sfox02]

I have two bed files, a file with peaks, and a file I created to be +/- 1kb of the TSS in mm9.

When I try to use bedtools intersect -v to find all of the peaks that are not overlapping with the 1kb region around a TSS, it returns the same peak file, but with one line different:

$ bedtools intersect -v -a ProE_replicatePeaks.bed -b 1kbTSS.bed > ATAC_Not_Pr.bed

$ wc -l ProE_replicatePeaks.bed

17440 ProE_replicatePeaks.bed

$ wc -l ATAC_Not_Pr.bed

17439 ATAC_Not_Pr.bed

I thought maybe I needed to sort the file containing the region around the TSS, but the same problem occurred.

Any input is appreciated! Thanks.

[Devon Ryan]

1. What version of bedtools?
2. Can you post a few relevant lines of each file?

Answer 1

[Alex Reynolds]

Are you expecting no overlaps? Perhaps try BEDOPS, in order to see what answer you get with an alternative toolkit.

First, sort the inputs:

$ sort-bed ProE_replicatePeaks.unknownSortOrder.bed > ProE_replicatePeaks.bed 
$ sort-bed 1kbTSS.unknownSortOrder.bed > 1kbTSS.bed

Then do the exclusion operation:

$ bedops --not-element-of 1 ProE_replicatePeaks.bed 1kbTSS.bed > ATAC_Not_Pr.bed

You could also do the opposite operation:

$ bedops --element-of 1 ProE_replicatePeaks.bed 1kbTSS.bed > ATAC_Pr.bed

In theory, if the first operation worked in the same way, you would get one BED element from the second operation, with which you can do a sanity check by comparing 1kbTSS.bed with ATAC_Pr.bed:

$ bedmap --echo --echo-map 1kbTSS.bed ATAC_Pr.bed > sanityCheck.bed