Does etraining overwrite previous models or refine them ?

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

Moderator: bioinf

Post Reply
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Does etraining overwrite previous models or refine them ?

Post by katharina »

Originally posted in the old forum by ebioman on 09.04.2014 - 11:55

Hello
Maybe this is plain obvious but I am a little bit confused and
just wanted to be on the safe side. I made a new species and
trained it, if I now start another "etraining" command with the
same species, will it start from scratch or actually refine it ?
THanks
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: Does etraining overwrite previous models or refine them ?

Post by katharina »

by Anand KS Rao on 23.07.2014 - 05:26
etraining over-writing
My understanding is that it is overwritten, and when you perform your accuracy testing, if it so happens that your sensitivity and/or specificity are reduced, you might want to revert to the better performing set of config files.
For this reason, it might be good to save your old config files
and / or
give a slightly different species name when you repeat your training, and you create config files for new species using new_species.pl, so that there is no chance of overlap, since these files are created in species specific folders.
I think my answer is correct, but will defer to Mario / Katharina's expertise.
Post Reply