intersection of two GFFs

Question

[BioDH]

I have two GFF files.

277(A.gff) and 1628(B.gff) regions were defined in each GFF.

What I would like to know is how many regions are in the intersection of two GFFs.

I used bedtools as following.

bedtoolls intersect -a A.gff -b B.gff -s -wa | wc -l

124

bedtoolls intersect -a B.gff -b A.gff -s -wa | wc -l

124

I though 124 entries were belonged to both.

But how can the followings be explained?

bedtools intersect -a A.gff -b B.gff -s -v  | wc -l

183

bedtoolls intersect -a B.gff -b A.gff -s -v | wc -l

1505

A.gff has 277, if A ∩ B = 124, A-B has to be 153. however, bedtools intersect -a A.gff -b B.gff -s -v | wc -l" tells 124

How can I get the intersection correctly?

[GenoMax]

Take a look at AGAT which is a proper GFF toolkit. You should be able to find overlapping features using one of the programs here.

[Juke34]

You can somehow use AGAT for that indeed but I'm not sure it is what @BioDH is looking for. You can use agat_sp_complement_annotations.pl, it will tell you how many features (e.g genes) you have in each annotation and then add in annotation A all genes from annotation B that do not overlap annotation A. You will have then a total of the new annotation A. Then you can deduce how many genes from B were overlapping A (But not the opposite). So you should then perform the opposite too, using B as reference and A as target.

Or maybe using agat_sp_merge_annotations.pl will tell you how many features were overlapping (I'm not remembering what information is provided by the tool).

[Juke34]

What do you mean exactly by regions? Could you show a sample of such region?