Understanding the number of intersection in bedtools jaccard

Question

[QLFblaireau]

Hello, I am using bedtools jaccard to compare two vcf files, as:

bedtools jaccard -a ancestors.calls.norm.snp.vcf.gz -b GC078310.calls.norm.snp.vcf.gz
intersection    union-intersection      jaccard n_intersections
1606899 1806667 0.889427        1536700

What I do not get is why n_intersections is equal to 1536700. Especially, the difference between intersection and n_intersections is not very clear to my mind.

Any help would be greatly appreciated.

Thanks a lot!

Answer 1

[i.sudbery]

If you have two 100bp regions that are 50% overlapped:

region 1 ----------
region 2      -----------
              |---|
               50bp

Here their are 50 bp overlapped so "intersection" is 50, and that is contained within 1 overlap (n_intersection = 1)

So "intersection" is the number of base pairs that are overlapped, while "n_intersections" is the number of overlaps between two intervals that those overlapped base pairs come from.