Gene prediction with UTR=on option

Discussions about predicting genes with AUGUSTUS. Not covered here: WebAUGUSTUS and BRAKER1

Moderator: bioinf

Post Reply
s.costagli3
Posts: 3
Joined: Sun Dec 12, 2021 5:19 pm

Gene prediction with UTR=on option

Post by s.costagli3 »

Good afternoon,
I am carrying out some gene predictions on my species of interest, using the AUGUSTUS parameters defined by training on WebAUGUSTUS.
By performing gene prediction with the utr=on option, the number of genes predicted is significantly reduced compared to the gene prediction with the utr=off option. In addition, when the UTR option is on, all predicted genes start at nucleotide 1 of the scaffolds into which the genome of interest is divided. I wanted to try to understand a possible explanation to these results.

Thank you very much, best regards,
Simone
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: Gene prediction with UTR=on option

Post by katharina »

Fully automated UTR training of AUGUSTUS sometimes works great, and sometimes produces really poor results, as in your case. To date, I have no automated way to "know" beforehand.

In some cases, the problem can be improved upon by setting the /UtrModel/polyasig_consensus in ${species}_parameters.cfg . You can also play with the pattern weight parameters in the same file.
Post Reply