AUGUSTUS Forum

Originally posted in the old forum by Clint Dalrymple on 01.02.2014 - 12:41

I've so far tried this with two different complete genomes, prepared with CEGMA 2.4.010312 and received the same error. This is on Scientific Linux 6.5, Augustus 3.0

[clint@genome test]$ autoAug.pl --Species=WPainted --genome=genome.fa --trainingset=augustus-training.gff --useexisting
2 checking for installed programs ... ok.
1 Checking fasta headers in file /home/clint/genomics/test/genome.fa...
2 All necessary directories have been created unter /home/clint/genomics/test/autoAug/trainingSet.
1 Using existing cDNA alignments and hints.
1 ####### Step 1: Training AUGUSTUS (no UTR models) #######
2 perl /home/clint/augustus-3.0/scripts/autoAugTrain.pl -t=/home/clint/genomics/test/augustus-training.gff -s=WPainted --useexisting -g=/home/clint/genomics/test/autoAug/seq/genome_clean.fa -w=/home/clint/genomics/test/autoAug -v -v --opt=1
2 Reusing existing training set training.gb.
2 Running command "perl /home/clint/augustus-3.0/config//../scripts/new_species.pl --species=WPainted --silent --ignore 1>new_species.stdout" to create AUGUSTUS config files for new species WPainted
2 Now training AUGUSTUS parameters for WPainted.
1 training.gb contains 378 sequences and 378 genes, each sequence contains 1 gene(s) on average.
1 test/evaluation set training.gb.test contains 37 sequences and 37 genes.
1 training set training.gb.train contains 341 sequences and 341 genes.
1 training.gb.train.test contains 300 sequences and 300 genes.
1 training.gb.onlytrain contains 41 sequences and 41 genes.
2 Seting value of "stopCodonExcludedFromCDS" in WPainted_parameters.cfg to "true"
2 First try with etraining: etraining --species=WPainted training.gb.train >train.out ...failed to execute:
failed to execute: No such file or directory
[clint@genome test]$ which etraining
~/augustus-3.0/bin/etraining
[clint@genome test]$

by katharina on 03.02.2014 - 12:03
I am not 100% sure, but it seems possible to me that Perl might not be able to handle the "tilde" (~) for home directory. Please try to export the full path to etraining, instead of the abbreviated path version.

by Clint Dalrymple on 05.02.2014 - 14:52
I'm not sure if there's some concatenation going on in the 'which' display or elsewhere, as the 'echo $PATH' has "...:/home/clint/augustus-3.0/bin/:/home/clint/augustus-3.0/scripts/:..."
Any other suggestions? I do suspect this could be at least partially my environment itself, as while it is a standard Red Hat derivative, I was having the same problem with the previous release of Augustus right before 3.0 came out.

by katharina on 06.02.2014 - 15:27
I don't have any experience with RedHat. We use Ubuntu most of the time. It seems really strange to me that autoAug.pl should not find a tool that is clearly in your path. You could try to add the full path to etraining to the autoAug-Perl scripts (if that helps, you'll probably have to add the paths to all tools that are called). Of course that's not how we intend users to use autoAug.pl, but it might work for you.

by Clint Dalrymple on 07.02.2014 - 20:12
I tried doing such and received the same error; I tried running etraining directly on the applicable file (training.gb.train) and received:
etraining: ERROR
Input file not in genbank format.
Additionally, after re-running, this is the content of the train.err for both genomes I have tried to work with.

by katharina on 10.02.2014 - 18:26
Please make the supposed genbank file available to me and e-mail the exact command line that you used to me (augustus-web[at]uni-greifswald.de).
Katharina

by JoseC on 10.11.2014 - 02:40
I am having the same error. I am using Ubuntu 14.04 and augustus 3.0.
1 ####### Step 1: Training AUGUSTUS (no UTR models) #######
Error: missing training file!
Any suggetsion