Remove overlapping features from a gtf file?

Question

[RoryC]

Hi, this seems like quite a straightforward question so apologies if it has been asked before (I couldn't find anything similar). I have a .gtf file containing CDS coordinates for a chromosome, and I plan to extract codons containing 4d sites. Therefore I would like to remove any CDS that overlap (on the same, or opposite, strand) so there is no ambiguity about what is a 4d site and what isn't. I've been trying to do this with bedtools but I'm not having much luck, as intersect would need to have an option where features that overlap 100% are ignored for me to compare the file to itself. Thanks

[dally]

Bedtools intersect -v option with -r 1 does not work?

[RoryC]

Hi, thanks for your answer. Do you mean -v -f 1 -r? This gives me zero output as every CDS is overlapped 100%.

Answer 1

[RoryC]

So after coming back to this some time after I think I've found a relatively simple way of doing this. First use bedtools merge with -o count to merge overlapping elements and add a fourth column which shows how many original elements contributed to the new elements. Then use a command prompt to remove any rows that have a number greater than 1 in the fourth column, thus removing anything that originally overlapped and was merged. For example with a bed file (the same could be done with gtf):

bedtools merge -i file.bed -c 1 -o count | awk ' { if($4==1) print $0} ' > newfile.bed

Answer 2

[GouthamAtla]

You can use the script (dexseq_prepare_annotation.py) given in the DEXSeq package to collapse overlapping exons. See Figure.1 of this paper http://genome.cshlp.org/content/22/10/2008.full

If this is not what you wanted, you may need to tweak the script a bit.