Some confusions about optimize_augustus.pl

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

Moderator: bioinf

Post Reply
Yafei
Posts: 11
Joined: Mon Nov 30, 2015 2:18 am

Some confusions about optimize_augustus.pl

Post by Yafei »

Hi guys,

I am using optimize_augustus.pl to optimize my 'new' species. But It has been taking over six days on 24 CPU =-= importantly, the time will be out soon because I set the max running time is 6 days and 23 hours. So I have two confusions here, could you help me figure out^^?

my command line looks like:perl optimize_augustus.pl --species=As As.genes.gb.train

1. I did not set the option --cpus=24, so if I would not set the cpus number, dose the script only use 1 cpu?

2. As I describe above, The time will be out soon and the optimize_augustus.pl won't be finished, so, do any commands or scripts could continue the unfinished optimize training? And also, I found the cue (After optimize_augustus.pl has finished or (after you have interrupted it) you should (re)train AUGUSTUS with the meta parameters it has set.) from http://www.molecularevolution.org/molev ... ining.html. does it mean that interrupting the optimize_augustus.pl is not a big deal, you could still use the interrupting results to do downstream analysis (etraining --species=As As.genes.gb.train) ?

Thanks a lot,

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

Re: Some confusions about optimize_augustus.pl

Post by katharina »

Yafei wrote: 1. I did not set the option --cpus=24, so if I would not set the cpus number, dose the script only use 1 cpu?
By default, only 1 CPU is used.
Yafei wrote: 2. As I describe above, The time will be out soon and the optimize_augustus.pl won't be finished, so, do any commands or scripts could continue the unfinished optimize training?
You can restart the original command. This will result in starting optimization with the paramters that have been optimized until the previous run was interrupted. This may be faster than starting from scratch.
Yafei wrote: And also, I found the cue (After optimize_augustus.pl has finished or (after you have interrupted it) you should (re)train AUGUSTUS with the meta parameters it has set.) from http://www.molecularevolution.org/molev ... ining.html. does it mean that interrupting the optimize_augustus.pl is not a big deal, you could still use the interrupting results to do downstream analysis (etraining --species=As As.genes.gb.train) ?
It is possible to interrupt optimize_augustus.pl before it finishes, by itself. However, this will mean that you will further use paramters optimized until the point at which you interrupted optimization. You could redirect the output of optimize_augustus.pl and in check these "log files" whether and how much accuracy is still improving. If you cannot see much of a change, anymore, it is less likely to make much of a difference if you wait or interrupt.

Katharina
Yafei
Posts: 11
Joined: Mon Nov 30, 2015 2:18 am

Re: Some confusions about optimize_augustus.pl

Post by Yafei »

Hi Katharina,

Thanks a lot ^^

Best,
Yafei
Post Reply