Training Augustus for Gene Prediction

Question

[shiv]

I am working on a plant genome and have a genome FASTA file. I also have GenBank file containing some manually curated reference gene models from a previous genome version. My species is not available in the Augustus species folder. How can I use both datasets (genome + genbank gene models) to train Augustus using augustus itself or may be using BUSCO and perform gene prediction effectively?

Thanks in advance!!

Answer 1

[colindaven]

GALBA can help you to train Augustus models on your genome https://github.com/Gaius-Augustus/GALBA?tab=readme-ov-file#what-is-galba

That said, the authors recommend BRAKER3 if you have RNA-seq.

[Buffo]

I've tested Braker2 using RNAseq on a non-model organism, and it does a terrible job. The main problem is the intron prediction.