Gene prediction - new species - augustus

[morko]

Hi,
I am new to the bioinformatics field and I already tried getting my gene prediction with busco+augustus, scipio(merged proteins from cryptodb)+augustus but the results weren't corresponding to the expectations and specifics of the organism - lots of introns.

The organism is cryptosporidium proliferans.
I've only got the assembly from Mira. Is it possible to use coding and transcript sequences from close species (cryptosporidium muris - cryptodb files) to get hints and then reliable gene prediction?

I am not sure how to train augustus right and then get good results. Or is there a better software you'd recommend?

Thanks.

[neavemj]

Hi Morko,

I would recommend MAKER (http://www.yandell-lab.org/software/maker.html). This is an annotation pipeline that will allow you to use homology-based predictions from closely related species, combined with computational prediction methods (such as augustus and GeneMark). The pipeline will run all of these methods then produce consensus predictions of the best gene models.

Cheers,

Matt.