How to merge gff files deriving from several evidences

Question

[marco.sollitto]

Hi, I'd like to understand which methods are useful for merging gff3 files deriving from different evidences.

When I run EvidenceModeler, it underestimates, and I have a BUSCO result lower than the single evidence.

Thanks,

Marco

[Juke34]

Whar do you mean by "I have a BUSCO result lower than the single evidence"?

[Mike]

hey Marco, what did you end up using? Id love to hear, thanks.

Answer 1

[Trivas]

I like the tools GffRead and GffCompare. The later can give you a lot of information when comparing different GFF or GTF files that might point you in the right direction.

[marco.sollitto]

I have tried both of them, but a lot of "duplicates" still remain. Through these tools, I reach a % of complete BUSCO ~93% (40% of "duplicates").

My genome has a ~98% of Complete BUSCOs%.

[Juke34]

When computing BUSCO on annotation you must remove isoforms to avoid to over-estimate "duplicates"

[marco.sollitto]

Yep, I know about different isoforms, but, in the file there are still several hits deriving from different evidences (with some nucleotides of difference, so no real isoforms, but only low accuracy in gff annotations).

[Juke34]

This is normal, most often annotations have lower BUSCO score than the assembly