Unexpected results with bedtools merge.

Question

[Coryza]

I am trying to merge a BED file containing all repeat-masked positions of a genome, however that gives me unexpected results.

For example I have a sorted unmerged BED file;

Nitab4.5_0000001        1       383
Nitab4.5_0000001        384     384
Nitab4.5_0000001        385     385
Nitab4.5_0000001        386     387
Nitab4.5_0000001        388     388
Nitab4.5_0000001        389     389
Nitab4.5_0000001        390     390
Nitab4.5_0000001        391     395
Nitab4.5_0000001        396     402
Nitab4.5_0000001        403     404

The merged BED file is;

Nitab4.5_0000001        1       395
Nitab4.5_0000001        396     402
Nitab4.5_0000001        403     404

I don't understand why this isn't a single feature? Furthermore I also get 0-based coordinates while these are not in the sorted BED file.

The sorted BED file;

Nitab4.5_0000003        1       1
Nitab4.5_0000003        2       2
Nitab4.5_0000003        3       4
Nitab4.5_0000003        5       9
Nitab4.5_0000003        10      11
Nitab4.5_0000003        12      16
Nitab4.5_0000003        17      24
Nitab4.5_0000003        25      28
Nitab4.5_0000003        29      73
Nitab4.5_0000003        74      90
......

The merged BED file;

Nitab4.5_0000003        0       4
Nitab4.5_0000003        5       9
Nitab4.5_0000003        10      11
Nitab4.5_0000003        12      16
Nitab4.5_0000003        17      24
Nitab4.5_0000003        25      28
Nitab4.5_0000003        29      73
Nitab4.5_0000003        74      90
Nitab4.5_0000003        91      213
Nitab4.5_0000003        214     221

Anyone who could help me with this?

Answer 1

[Alex Reynolds]

BED elements are usually 0-indexed, half-open, so these elements in your input are not strictly BED:

Nitab4.5_0000001        384     384
Nitab4.5_0000001        385     385
...
Nitab4.5_0000001        388     388
Nitab4.5_0000001        389     389
Nitab4.5_0000001        390     390

A half-open interval means up to but not including, and is sometimes written as [a, b), where a < b. Instead, you may want to use [a-1, a) to represent those elements, before doing set operations.