UTR-training decreased the 'sensitivity' and 'specificity'

Discussions about training AUGUSTUS from various sources of evidence. Not discussed here: BRAKER1 and WebAUGUSTUS!

Moderator: bioinf

Post Reply
LP_Lv
Posts: 2
Joined: Thu Sep 10, 2020 4:58 pm

UTR-training decreased the 'sensitivity' and 'specificity'

Post by LP_Lv »

Hi,
I have a question as described in the topic. I'm trying to train Augustus to predict genes of a non-model species.
I followed the instructions at here: http://bioinf.uni-greifswald.de/augustu ... ining.html .
The sensitivity of genes is 0.66 and specificity is 0.495 after optimizing for 5 rounds.
However, when I try to triain the UTR parameters with about 500 bothUTR-genes as described in http://bioinf.uni-greifswald.de/bioinf/ ... TRTraining , the sensitivity of genes decreased to 0.267 (specificity is 0.267 too) after running the "optimize_augustus.pl".
Did I do something wrong or miss some options ?

Thank you in advance,
Liping
LP_Lv
Posts: 2
Joined: Thu Sep 10, 2020 4:58 pm

Re: UTR-training decreased the 'sensitivity' and 'specificity'

Post by LP_Lv »

well, I know the reason now.
There is a sentence at the instruction :
The training gene structure file must not contain UTR examples for all genes, but for a sufficiently high number, UTRs must be present
But I missed that.
Post Reply