Hi,
is there a way to restrict the behaviour of bedtools to a line-by-line comparison?
To be more clear, if I have 2 BED files:
File_A:
chr1 1 100
chr2 1 100
and
File_B:
chr2 50 70
chr3 1 100
I'd like bedtools subtract to eliminate from File_A the interval contained in one line only if the interval in the corresponding line in file_B overlap. Therefore, in the present example no intervals should be subtracted from File_A, but I think by default bedtools is scanning one interval of File_A against all other intervals of File_B.
Is this possible?
Thank you in advance!
1 answer
To do this with BEDOPS bedops --difference and bash:
$ while read A <&3 && read B <&4; do bedops --difference <(echo -e "${A}") <(echo -e "${B}"); done 3<A.bed 4<B.bed
Given:
$ cat A.bed
chr1 1 100
chr2 1 100
$ cat B.bed
chr2 50 70
chr3 1 100
This gives the following result:
$ while read A <&3 && read B <&4; do bedops --difference <(echo -e "${A}") <(echo -e "${B}"); done 3<A.bed 4<B.bed
chr1 1 100
chr2 1 100
To confirm the line-by-line comparison, you could replace bedops --difference with paste:
$ while read A <&3 && read B <&4; do paste <(echo -e "${A}") <(echo -e "${B}"); done 3<A.bed 4<B.bed
chr1 1 100 chr2 50 70
chr2 1 100 chr3 1 100
Note that bedops --difference is directional, i.e. it depends upon the ordering of inputs. For instance:
$ while read A <&3 && read B <&4; do bedops --difference <(echo -e "${B}") <(echo -e "${A}"); done 3<A.bed 4<B.bed
chr2 50 70
chr3 1 100
Of course, to make this work reliably, you will first want to verify that both A.bed and B.bed have the same number of lines before you do any of this. You could, for example, test and compare with wc -l and a numerical comparison test operation in bash.
Log in to answer this question.
Thanks a lot Alex, I'll try this out. Cheers!