JIGSAW gene prediction combiner - use for prokaryotes?!

[all_your_base]

Hi, I have a recently assembled bacterial genome. It consists of a a couple dozens scaffolds, and I have put them through three gene prediction methods: Prodigal, Glimmer, and GeneMark.

Now that I have three different gtf files describing putative genes, I want to somehow combine these results into a consensus list. I have been looking into such a combiner called JIGSAW and was wondering if anyone has experience using this software with bacteria (I think it was made with eukarylotes in mind).

I am running with the following command:

jigsaw -l -f "myGenome.fasta" -m "jigsaw.output" -e "myEvidenceFile"

And my evidence file looks like this:

scaffolds_GeneMark.gff gff geneprediction coding 1.0
scaffolds_Prodigal.gff gff geneprediction coding 1.0
scaffolds_Glimmer.gff gff geneprediction coding 1.0

Typically, JIGSAW wants the user to provide the type of exon that was annotated (start, internal, end, etc), but since this is a prokaryote, I thought maybe it was best just to use the "coding" identifier.

The problem is, for a particular contig that has an average of 10 genes predicted from each individual method (7 of which overlap perfectly between all three methods), JIGSAW is only predicting 2 genes!

Any comments or suggestions and much appreciated

[Inma]

I did the same. I predicted ORFs in a complete bacteria genome by glimmer, genemark and prodigal. However, how I can integrate all this information and decide which ORFs are the most probables ones and which ones are not?

[Loddi]

Look on Pubmed for "Reganor", I remember this software does exactly this combination of different gene prediction tools.

Best,
Loddi