Identify reads overlapping a junction in gff file

[jazz]

Dear all,

I have a gff annotation file with coordinates for an overlapping genomic feature, e.g.
chr1 1-100
chr1 100-250
chr1 250-300
chr1 300-500

and an alignment file (bam/bed) with reads mapping to that chromosome. I need to extract reads that are specifically overlapping the junctions of that genomic feature. So, in a bed file like this,

chr1 50-70
chr1 90-110
chr1 150-170
chr1 245-255

Only extract,
chr1 90-110
chr1 245-255

Any existing tools that can help me do that. I am not well versed in bioinformatics so I can't write my own scripts but am able to run existing tools.

Any ideas would be appreciated.

Thanks

[bruce01]

BEdtools intersect will do this, you need tab-separated input so:

sed 's/\-/\t/g' <gff_or_bed> | sed 's/\s */\t/g' > <new-gff_or_bed>

should work for that

[jazz]

Thanks for the prompt reply bruce01, but I thought that bedtools will report all the common overlap features between my gff and bam/bed. I only need to extract the reads which are overlapping the junctions of my genomic feature. So, discard any reads which are completely within the coordinates of a single feature.
Am I missing something here?

[bruce01]

Oh, you want reads to sit in two concurrent features, I misread and thought you just wanted to be within one feature. No tools I know to do that, but thats because I never wanted to do that.

What are you trying to find, fusion genes?