joingenes segmentation fault

[ChuShin]

Hello,
We ran into a segmentation fault error in joingenes while running braker pipeline(v2.1.5, augustus 3.3.3). I extracted the first offending lines (see below a.gtf) and I can reproduce the error in terminal (see log below). Please can you help?

Thanks.

joingenes --genesets=a.gtf,a.gtf --priorities=2,1 --output=output.gtf
# After loading a.gtf (Priority 2) there are 2 transcripts in transcript list.
# After loading a.gtf (Priority 1) there are 4 transcripts in transcript list.
Segmentation fault

and a.gtf attached below:

Scaffold01577 AUGUSTUS gene 26249 28343 0.8 + . g160845
Scaffold01577 AUGUSTUS transcript 26249 28343 0.44 + . g160845.t1
Scaffold01577 AUGUSTUS start_codon 26249 26251 . + 0 transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS CDS 26249 27341 1 + 0 transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS exon 26249 27341 . + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS intron 27342 27405 0.49 + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS CDS 27406 27929 0.48 + 2 transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS exon 27406 27929 . + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS intron 27930 28004 0.99 + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS CDS 28005 28112 0.99 + 0 transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS exon 28005 28112 . + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS intron 28113 28189 0.99 + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS CDS 28190 28290 0.9 + 0 transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS exon 28190 28290 . + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS intron 28291 28343 0.9 + . transcript_id "g160845.t1"; gene_id "g160845";
Scaffold01577 AUGUSTUS transcript 26249 27983 0.36 + . g160845.t2
Scaffold01577 AUGUSTUS start_codon 26249 26251 . + 0 transcript_id "g160845.t2"; gene_id "g160845";
Scaffold01577 AUGUSTUS CDS 26249 27341 1 + 0 transcript_id "g160845.t2"; gene_id "g160845";
Scaffold01577 AUGUSTUS exon 26249 27341 . + . transcript_id "g160845.t2"; gene_id "g160845";
Scaffold01577 AUGUSTUS intron 27342 27405 0.5 + . transcript_id "g160845.t2"; gene_id "g160845";
Scaffold01577 AUGUSTUS CDS 27406 27983 0.36 + 2 transcript_id "g160845.t2"; gene_id "g160845";
Scaffold01577 AUGUSTUS exon 27406 27983 . + . transcript_id "g160845.t2"; gene_id "g160845";
Scaffold01577 AUGUSTUS stop_codon 27981 27983 . + 0 transcript_id "g160845.t2"; gene_id "g160845";

[tomasbruna]

Hello @ChuShin,

I tried to reproduce the error on 3 different systems, unfortunately, I could not reproduce the segmentation fault, the binary worked correctly each time.

Does it fail like this on all inputs you supply? Can you try to re-compile joingenes, or run it on a different machine?

Tomas

[ricardo-aaron]

I also get that error:
Thu Aug 13 15:35:07 2020: Log information is stored in file /elpool/rchavez/myspecies/sello/braker/braker_run1/braker.log
ERROR in file /elpool/software/BRAKER/scripts/braker.pl at line 9472
Failed to execute: /elpool/software/Augustus/bin/joingenes --genesets=/elpool/rchavez/myspecies/sello/braker/braker_run1/augustus.E.gtf,/elpool/rchavez/myspecies/sello/braker/braker_run1/augustus.Ppri5.gtf_filtered --priorities=2,1 --output=/elpool/rchavez/myspecies/sello/braker/braker_run1/join.gtf 1> /dev/null 2> /elpool/rchavez/myspecies/sello/braker/braker_run1/errors/joingenes.err!

joingenes.err contains:
Segmentation fault (core dumped)

this is the output of the joingenes command:
After loading /elpool/rchavez/myspecies/sello/braker/braker_run1/augustus.E.gtf (Priority 2) there are 24205 transcripts in transcript list.
After loading /elpool/rchavez/myspecies/sello/braker/braker_run1/augustus.Ppri5.gtf_filtered (Priority 1) there are 24205 transcripts in transcript list.
Segmentation fault (core dumped)

This with an Augustus git cloned today.

[KatharinaHoff]

@ChuShin thank you very much for creating a toy example data set. I am able to reproduce the Segmentation fault. Someone in our group will eventually look into this.
Katharina

[MarioStanke]

It is now an issue in the Augustus repository as joingenes belongs there:
Gaius-Augustus/Augustus#182

[bioinfowheat]

I'm also having this problem with the latest version of BRAKER2

error on run

[Thu Sep 17 15:49:26 2020] Enqueueing pair 329050/329379 (99.9%). Est. time left: 00:00:02 (hh:mm:ss)
[Thu Sep 17 15:50:28 2020] 329379/329379 (100%) pairs aligned
[Thu Sep 17 15:50:28 2020] Alignment of pairs finished
[Thu Sep 17 15:50:28 2020] Translating coordinates from local pair level to contig level
[Thu Sep 17 15:50:50 2020] Finished spliced alignment
[Thu Sep 17 15:50:51 2020] Processing the output
[Thu Sep 17 15:52:35 2020] Output processed
[Thu Sep 17 15:52:35 2020] ProtHint finished.
ERROR in file /data/programs/BRAKER2_v2.1.5/scripts/braker.pl at line 9481
Failed to execute: /data/programs/Augustus_v3.3.3/bin/joingenes --genesets=/mnt/griffin/marcel/Pca_genome/braker2_test/braker/augustus.E.gtf,/mnt/griffin/marcel/Pca_genome/braker2_test/braker/augustus.Ppri5.gtf_filtered --priorities=2,1 --output=/mnt/griffin/marcel/Pca_genome/braker2_test/braker/join.gtf 1> /dev/null 2> /mnt/griffin/marcel/Pca_genome/braker2_test/braker/errors/joingenes.err!

OK, check the error file

more /mnt/griffin/marcel/Pca_genome/braker2_test/braker/errors/joingenes.err
Load warning: Did not expect feature "internal". Known features are "CDS", "UTR", "3'-UTR", "5'-UTR", "exon", "intron", "gene", "transcript", "tss", "tts", "s
tart_codon" and "stop_codon". This feature is going to be ignored.
This warning may affect the result.
Load warning: Did not expect feature "initial". Known features are "CDS", "UTR", "3'-UTR", "5'-UTR", "exon", "intron", "gene", "transcript", "tss", "tts", "st
art_codon" and "stop_codon". This feature is going to be ignored.
This warning may affect the result.
Load warning: Did not expect feature "terminal". Known features are "CDS", "UTR", "3'-UTR", "5'-UTR", "exon", "intron", "gene", "transcript", "tss", "tts", "s
tart_codon" and "stop_codon". This feature is going to be ignored.
This warning may affect the result.
Segmentation fault

OK, so I am attempting to run this

/data/programs/Augustus_v3.3.3/bin/joingenes --genesets=/mnt/griffin/marcel/Pca_genome/braker2_test/braker/augustus.E.gtf,/mnt/griffin/marcel/Pca_genome/braker2_test/braker/augustus.Ppri5.gtf_filtered --priorities=2,1 --output=/mnt/griffin/marcel/Pca_genome/braker2_test/braker/join.gtf 1> /dev/null 2> /mnt/griffin/marcel/Pca_genome/braker2_test/braker/errors/joingenes.err

returns

Segmentation fault

[JudithR]

Hi,
Not sure whether this should go with BRAKER or Augustus. I posted it on Augustus as issue Gaius-Augustus/Augustus#260.
Joingenes clearly is part of Augustus, but the 'faulty' input is generated by BRAKER.

I'm running BRAKER with --etpmode
braker.pl --species=mySpecies --genome=myGenome_softmasked.fa --softmasking --bam=myGenome_STARAligned.sortedByCoord.out.noMito.fwd.bam,myGenome_STARAligned.sortedByCoord.out.noMito.rev.bam --UTR=on --cores=10 --stranded=+,- --prothints=prothint_augustus.gff --hints=hintsfile.gff --etpmode

The hintsfile.gff is from a previous BRAKER run (RNAseq only) without etpmode and OrthoDB, but otherwise identical.

I encountered issue Gaius-Augustus/Augustus#182 and subsequently upgraded Augustus to the latest git-checkout. Unfortunately the seqfault still occurs.
The offending joingenes command is:
joingenes --genesets=braker/augustus.E.gtf,braker/augustus.Ppri5.gtf_filtered --priorities=2,1 --output=braker/join.gtf 1> /dev/null 2> braker/errors/joingenes.err

This issue was only present in BRAKER 2.5, but is resolved in the latest BRAKER checkout.

I reran the job with both the latest BRAKER and Augustus checkouts.

There a similar issue seems to occur in a different step and not the same as mentioned above:
ERROR in file BRAKER/scripts/braker.pl at line 9760

This is the tail of the logfile

# File braker/augustus.Ppri5.gtf has 31546 supported transcripts, braker/augustus.E.gtf has 42337 supported transcripts
# Filtering those genes that have evidence by src=P from braker/augustus.Ppri5.gtf...
perl filter_augustus_gff.pl --in=braker/augustus.Ppri5.gff --src=P --out=braker/augustus.Ppri5.gtf_filtered
joingenes --genesets=braker/augustus.E.gtf,braker/augustus.Ppri5.gtf_filtered --priorities=2,1 --output=braker/join.gtf 1> /dev/null 2> /local_scratch/risse001/Taraxacum/braker/braker/errors/joingenes.err

The joingenes error log only gives warnings about unknown features prior to the segfault.

Load warning: Did not expect feature "terminal". Known features are "CDS", "UTR", "3'-UTR", "5'-UTR", "exon", "intron", "gene", "transcript", "tss", "tts", "start_codon" and "stop_codon". This feature is going to be ignored.
This warning may affect the result.
Load warning: Did not expect feature "internal". Known features are "CDS", "UTR", "3'-UTR", "5'-UTR", "exon", "intron", "gene", "transcript", "tss", "tts", "start_codon" and "stop_codon". This feature is going to be ignored.
This warning may affect the result.
Load warning: Did not expect feature "initial". Known features are "CDS", "UTR", "3'-UTR", "5'-UTR", "exon", "intron", "gene", "transcript", "tss", "tts", "start_codon" and "stop_codon". This feature is going to be ignored.
This warning may affect the result.

But this time both input files do contain data..

I'm not sure how to quickly fix this
If you could help, that would be greatly appreciated. Also what would be the fastest way to continue the BRAKER run?
Regards
Judith

[JudithR]

No longer relevant. The latest git checkout from Augustus does not give the error anymore. Somehow joingenes did not get recompiled with the latest checkout.
Apologies