bedtools intersect with match on name column

Question: bedtools intersect with match on name column

6 months ago by

corend • 70

corend • 70 wrote:

I have files a.bed and b.bed

a.bed

chr1 10 20 geneA
chr1 30 40 geneB

b.bed

chr1 5 15 geneA
chr1 10 25 geneB

if I intersectbed those files I will obtain:

chr1 10 15 geneA
chr1 10 20 geneA

But I would like the name column to match between files so that i will obtain:

chr1 10 15 geneA

If name column does not match, I don't want to output the intersection. I can't find any option in intersectbed to "force identical name", any suggestions?

intersectbed bedtools • 387 views

ADD COMMENT • link •

modified 6 months ago • written 6 months ago by corend • 70

6 months ago by

ATpoint ♦ 18k

Germany

ATpoint ♦ 18k wrote:

I would do a two-step process:

## First use the -wa -wb options to report both the entire entry of -a and of -b in case of overlap,
## and then use awk to check if the genes are the same.
## If so, keep the row, if not discard:

bedtools intersect -a a.txt -b b.txt -wa -wb | awk 'OFS="\t" {if ($4 == $8) print}' > intermediate.txt

## Then use intersect again to get the actual overlap:
bedtools intersect -a <(cut -f1-4 intermediate.txt) -b <(cut -f 5-8 intermediate.txt)

ADD COMMENT • link modified 6 months ago • written 6 months ago by ATpoint ♦ 18k

Sorry, it is not working, here is the problem:

In the second step, you don't intersect line by line, so the problem is still the same. Just try it and you will see.

ADD REPLY • link written 6 months ago by corend • 70

Based on the data you provided it gives exactly the result you intended:

enter image description here

ADD REPLY • link written 6 months ago by ATpoint ♦ 18k

My bad, my example was not a MWE, I will validate your answer. If you replace 10 25 by 10 35 here you will see the problem.

ADD REPLY • link written 6 months ago by corend • 70

6 months ago by

corend • 70

corend • 70 wrote:

I finaly found out how to do it:

As I want a Chromosome and name match, I just make a bedfile in the format chr+name start end name, and it works well.

ADD COMMENT • link written 6 months ago by corend • 70

Please log in to add an answer.

Similar posts • Search »