v1.0.2 augustus issue

[Sebastien-Raguideau]

When running ufcg (v1.0.2) profile, I get this error message:
ERROR! Failed subcommand :augustus --optCfgFile=/mnt/gpfs/seb/Applications/UFCG/config/ppx.cfg --predictionStart=24228 --predictionEnd=44228 --proteinprofile=/mnt/gpfs/seb/Applications/UFCG/config/model/pro/RPB2.hmm /home/sebr/seb/Database/Fungus_DB/UFCG/tmp/Lobtra1/tmp/Lobtra1/Lobtra1_scaffold_92.fna > /home/sebr/seb/Database/Fungus_DB/UFCG/tmp/Lobtra1/tmp/Lobtra1/Lobtra1_scaffold_92_p34228_RPB2.gff

I checked and this same genome dealt with normaly in ufcg v1.0.1. I have now to have ufcg (v1.0.1) for profiling and ufcg (v1.0.2) for building tree (it does not work in v1.0.1).

[endixk]

Could you please give me extra information by answering following questions:

• Does the error occurs on [all of / some of / one of] your genomes?
• From the run that produced the error, was there any successful augustus run before the error? (you can check it by running the same command with -v flag)
• Which error message did v1.0.1 tree module produce?

[Sebastien-Raguideau]

I just realised that I need to change my ufcg env to upgrade augustus. I didn't test it yet and will tel you if error persist after update.
Regarding v1.0.1 tree module issue, it was hanging for more than 1 hour without doing anything. Keyboard interrupt would unfreeze it and also exit it. For instance if I made a mistake in argument, it would not tell me, just hang and then after ctrl+c would print a message about it and exit. Same thing when calling ufcg with correct argument.

[Sebastien-Raguideau]

So I upgraded augustus to 3.5.0, the issue persists. To answer you question, yes there are successful augustus run before the failure.
No problem occurs when using v1.0.1.
I joined one of the genome for which annotation fail.
Lobtra1.fasta.gz

[endixk]

Thanks to your example file, I realized that the FASTA parsing algorithm is flawed 😅

The program is supposed to extract the FASTA entries with identical names detected by fastBlockSearch, but currently the code uses substring matching for some reason. If the FASTA headers have common prefix as your example, the program gets confused and fails to extract the precise sequence blocks, which eventually ends up with the error.

This behavior will soon be fixed in the next release.

[endixk]

This issue should be fixed from the v1.0.3 release. Please let me know if this persists.

[AdamVS]

I have also had this issue with some files passing in v1.0.5, and others not. It was fixed by sorting and renaming the scaffolds. Possibly also by setting a minimum length