ERROR in: Now looping and iteratively modify the GFA until all reads align end-to-end on the assembly graph

[octpalacios]

Hi, I’m using Hairsplitter on a large repetitive genome assembled with Flye, but I encountered the following error:

==== Now looping and iteratively modify the GFA until all reads align end-to-end on the assembly graph ====

Loop iteration 0...

• Going through the gaf file and listing the reads that do not align end-to-end on the assembly graph...
• Outputting non-duplexed reads in /crex/proj/snic2021-6-151/nobackup/private/Octavio/V.viatia.P24_XY_male_assembly/PacBio_reads/Flye_assembly/Hairsplitter/hairsplitter_out/tmp/reads.fa, discarding in total 87225 reads
• Pooling the reads that display similar behaviour on the assembly graph...
• Found 9726 solid bridges and 41160 solid piers.
• Computing the exact location of new links in the GFA and gap-filling if necessary... terminate called after throwing an instance of 'std::out_of_range' what(): basic_string::substr: __pos (which is 3792) > this->size() (which is 60)

Do you have any suggestions on how to fix this? Thanks in advance.

[RolandFaure]

Hi,

This is clearly a bug, which may originate from the very repetitive nature of your genome. I need to correct this. If you want a workaround, you can skip step 0 of the pipeline by toggling the --correct-assembly flag, but this will diminish the quality of the result.

To solve this bug, could you show me your command line ? What version of HairSplitter are you using ? And would it be possible to share the files so that I can take a look, or are the files confidential or too big ?

[octpalacios]

Hi,

I'm currently using HairSplitter v1.9.10, installed via Conda. I'm about to try updating it. I'm already using the "--correct-assembly" flag. Here’s the command I'm running:

hairsplitter.py -f $READS -i $ASM -x pacbio --correct-assembly -t 16 -o $OUT --clean --resume

Which files do you need? The GFA assembly and the reads? I can possibly share them via email if you keep them confidential. Thanks in advance.

[RolandFaure]

Yes, the GFA and the reads would be perfect for me, at roland.faure@irisa.fr. I will delete all the files once the debugging is done. Not using --correct-assembly will skip over the buggy part but will decrease the quality of the results.

[octpalacios]

Hi Roland,

The files are quite large, and I’m working on a way to send them. In the meantime, I’ll try running Hairsplitter without the "--correct-assembly" flag.

Thanks.

[RolandFaure]

Hi,

If this help, the qualities of the reads in the fastq file are not used, hence a fasta file of the reads is enough.

Roland

De: "Octavio M. Palacios-Gimenez" @.> À: "RolandFaure/Hairsplitter" @.> Cc: "roland faure" @.>, "Comment" @.> Envoyé: Mercredi 30 Octobre 2024 13:48:09 Objet: Re: [RolandFaure/Hairsplitter] ERROR in: Now looping and iteratively modify the GFA until all reads align end-to-end on the assembly graph (Issue

10)

Hi Roland,

The files are quite large, and I’m working on a way to send them. In the meantime, I’ll try running Hairsplitter without the "--correct-assembly" flag.

Thanks.

— Reply to this email directly, [ https://github.com/RolandFaure/Hairsplitter/issues/10#issuecomment-2447035877 | view it on GitHub ] , or [ https://github.com/notifications/unsubscribe-auth/APJJHA2RYAORYKQJB4MKYYLZ6DIQTAVCNFSM6AAAAABQ3PXKHCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINBXGAZTKOBXG4 | unsubscribe ] . You are receiving this because you commented. Message ID: < @.*** >