I can't run augustus

Discussions about compiler errors and installation problems

Moderator: bioinf

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

I can't run augustus

Post by katharina »

Originally posted in the old forum by Estanislao on 15.01.2013 - 01:29
I followed the 3 steps of instalation but the shell can´t recognized the comand augustus, please, can you tellme how I how to install augustus and run it??? I am using UBUNTU 12.04 ( I compiled the src yet)
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 15.01.2013 - 11:13

If I understand correctly, you have changed to the src directory:

Code: Select all

cd src
and run make:

Code: Select all

make
and set the environment variable AUGUSTUS_CONFIG_PATH:

Code: Select all

export AUGUSTUS_CONFIG_PATH=/my_path_to_AUGUSTUS/augustus/config/
And you are now still in the src directory?
If that's the case, your Linux system does not know where to find the augustus executable. There are at least three very general, not augustus-specific ways to solve this problem:

A) call the executable with its full path. Example: I installed augustus in /home/katharina/augustus, the executable then automatically resides in the folder /home/katharina/augustus/bin. My call would then be:
/home/katharina/augustus/bin/augustus ${parameters}
where ${parameters} are all the parameters that I want to use for running augustus.

B) change to the directory where the augustus binary exectuable resides and execute the local binary file (assuming the setting from example A):
cd /home/katharina/augustus/bin
./augustus ${parameters}

C) If you do not always want to specify the full path or to change to the augustus binary directory, you give information to your system where it should look for software executables. Your system will look for such information for a single user of your system in files like
~/.bashrc
~/.profile
(The ~/ stands for /home/yourusername, so in my case, it would stand for /home/katharina.)
Open for example the file ~/.bashrc with an editor of your choice. Scroll to the bottom of the file. If there already an entry like
PATH=/something:/something/else:$PATH
then add the path to your augustus binary. In my example from A, I would add:
PATH=/something:/something/else:/home/katharina/augustus/bin/:
/home/katharina/augustus/scripts/:$PATH
If there is no PATH statement made, simply create it from scratch, for example I would write:
PATH=/home/katharina/augustus/bin/:/home/katharina/augustus/scripts/:$PATH
You can also add the export of the AUGUSTUS_CONFIG_PATH to those files!
Then save and close the file. If you open a new terminal window, the system will immediately find the augustus executable. If you want to continue working in a terminal window that was already open before you modifed the ~/.bashrc or ~/.profile, you have to load the file, first, e.g.:
source ~/.bashrc

For further reading, I recommend http://wiki.ubuntuusers.de/Bash/bashrc
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 17.01.2013 - 18:08

Hi Estanislao,
the job that you link to returned two important files to you: AutoAug.log and AutoAug.err
From your job's AutoAug.log file, I quote:
"1 - WARNING: Detected whitespace in fasta header of file /data/www/augtrain/webdata/trainecfF2B5n/genome.fa. This may later on cause problems! If the pipeline turns out to crash, please clean up the fasta headers, e.g. by using simplifyFastaHeaders.pl. This message will be suppressed from now on!
1 - WARNING: Detected whitespace in fasta header of file /data/www/augtrain/webdata/trainecfF2B5n/protein.fa. This may later on cause problems! If the pipeline turns out to crash, please clean up the fasta headers, e.g. by using simplifyFastaHeaders.pl. This message will be suppressed from now on!
1 - WARNING: Fasta headers in file /data/www/augtrain/webdata/trainecfF2B5n/protein.fa seem to contain non-letter and non-number characters. That means they may contain some kind of special character. This may later on cause problems! If the pipeline turns out to crash, please clean up the fasta headers, e.g. by using simplifyFastaHeaders.pl. This message will be suppressed from now on!"
The first source of your problem is therefore that your file contains whitespaces in places where they should not be. Concerning your file in particular, there are lots of empty newlines between the fasta entries, and the main problem is that the file starts with empty newlines instead of the first fasta headers. Please remove all empty newlines, and particularly the empty newlines before the first fasta entry and try again.
Further, I quote from the AutoAug.err file:
"Number of lines in gff file produced by scipio was zero! This is likely to cause problems because no training gene genbank entries can be created! etraining will crash when the training gene genbank file is empty!
ERROR: training.gb is empty. Possible reasons:
a) features in a provided training gene structure gff file were not compliant with the autoAug.pl pipeline (for instructions read at e.g. http://bioinf.uni-greifswald.de/webaugu ... #structure
b) Scipio failed to generate training gene structures
This will cause a crash of the autoAug.pl pipeline!
failed to execute: No such file or directory"
There are two possible reasons. The first reason (a) is unlikely to be the case for you, since you did not submit a manually created training gene structure file. But reason (b) is exactly what lead to the crash of the pipeline: Scipio was not able to generate any training gene structures from the provided data. It is theoretically possible that it was "really impossible" to create training gene structures but I strongly believe that there will be at least some few structures generated if you solve the empty newline problem.
Katharina
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 17.01.2013 - 18:11
I would like to add the following: you tried to run AUGUSTUS gene prediction jobs with parameter sets that have not been trained, successfully, i.e. you supplied training job IDs to the prediction interface for jobs that reported a severe Error message. This will never work! It is not possible to predict genes with parameter sets that have not been created successfully!
Edit:
I see that you are still trying the impossible. I give you an example: You job with the ID trainecfF2B5n failed! Parameters were NOT trained by this job. It is absolutely pointless to submit jobs via the prediction interface specifying "trainecfF2B5n" as a parameter ID.
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by Estanislao on 17.01.2013 - 19:12
Dear katharina:
I can´t training augustus even i solved the the empty newline problem:
http://bioinf.uni-greifswald.de/webaugu ... a6be7f0080
this is the protein training file without
https://www.dropbox.com/s/bghzvf8kv4xur ... ining.fa?m
Why I can´t do thi?
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by Estanislao on 17.01.2013 - 19:22
pd: the proteiins are the diferents but closely related worms (the same genus)
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 17.01.2013 - 19:23
Dear Estanislao,
the AutoAug.err file of the job in question reports to you:
"Number of training genes is with 10 too low (at least 100 genes required)! Training aborted.
failed to execute: No such file or directory"
It precisely means what it says: the provided data was not sufficient for generating number of training gene structures that is higher than 100.
In general, it is not recommended to rely on the results of an AUGUSTUS training if it was trained on very few genes. So even if you had 101 training gene structures, and the job did not report any error, you should inspect the prediction results with such a parameter set very carefully before trusting them.
In summary: you cannot train AUGUSTUS with the data that you submitted to our web service because it was not possible to generate a sufficient number of training gene structures form protein and genome file using Scipio.
There may be other options to get training gene structures for your species, but it is out of scope to explain them here. I encourage you to read our Wiki pages and do some research on other gene prediction tools.
Katharina
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by Estanislao on 17.01.2013 - 19:44
Dear Katherina:
Can I run augustus with C elegans parameters? Is It the same if I training augustus for my species with C elegas data, because this organisms is related and I don´t have more information (proteins, genes, or cdna) for my specie.
Thank you very much!
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by Estanislao on 17.01.2013 - 19:59
PD: Can I use for trainig putative proteins?, my specieis a worm like C elegans, Meloidogyne incognita.
Can I use the proteins data of C elegans for trainign augustos for M incognita?
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 17.01.2013 - 20:18
I will not answer to your question specifically because it is impossible to know the answer without some serious data analysis.
In general, it should not have a huge impact whether a gene is labelled as "putative" or something else. What matters is that an alignment between the protein sequences and the target genome sequence will in combination with Scipio produce a valid gene structure. A valid gene structure cannot have gaps, or in-frame stop-codons, or other problems that will make it an invalid gene structure.
Second, the quality of a genome assembly has a huge impact on generating training gene structures. If the assembly is too fragmented, it will be next to impossible to use it for training AUGUSTUS.
It is up to your personal expertise and the results of your personal experiments, whether it will be possible to train AUGUSTUS using the data form species that you mentioned.
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by Estanislao on 17.01.2013 - 23:42
My specie have more than 9000 contigs, but in Its papers of secuencing the autors told that they could trainig a software (eugene) for that specie
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 18.01.2013 - 10:05
As I said before, there are other options to generate training gene structures.
If a training gene set or an annotation already exists, I suggest that you reformat it to the requirements of the autoAug pipeline (submission of training gene structure file) and train AUGUSTUS with those gene structures.
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by madhu on 29.11.2013 - 13:14

hello am trying to run Augustus its giving error Environment variable AUGUSTUS_CONFIG_PATH not set then i went and tried to give config file path at the script directory new_species.pl still it shows the error what can be done?
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by katharina on 29.11.2013 - 15:06
Let's assume you downloaded and unpacked augustus to a directory /home/user/, i.e. all sources associated with augustus are located at /home/user/augustus, then you need to specify:
AUGUSTUS_CONFIG_PATH=/home/user/augustus/config
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by madhu on 16.12.2013 - 10:03
hello optimize_augustus.pl is showing error that it cannot find the executable directory , when i run in the command line i have given the path of etraining still it shows error
User avatar
katharina
Site Admin
Posts: 531
Joined: Wed Nov 18, 2015 6:14 pm
Location: Greifswald
Contact:

Re: I can't run augustus

Post by katharina »

by k on 16.12.2013 - 17:54
Please read the second post in this thread and repeat all steps listed there, carefully. This will definitely work for a Linux system.
Post Reply