Segfault from latest version

[jetrz]

Hello, when trying to use the latest --ont feature, I am getting a segfault.

I am trying to perform diploid and haploid assembly respectively on:

1. HG002: https://s3-us-west-2.amazonaws.com/human-pangenomics/working/HPRC_PLUS/HG002/raw_data/nanopore/HG002_ONT_PAD64459_Guppy_3.2.fastq.gz
2. Fruit Fly: https://www.ncbi.nlm.nih.gov/sra/SRR23215007

Are these datasets not suitable? Thank you.

[chhylp123]

The public data we tested including: HG002 from paoloshasta/shasta#28 and a tomato from https://obj.umiacs.umd.edu/marbl_publications/duplex/Solanum_lycopersicum_heinz1706/UL/R10.4_40x.noduplex.fastq.gz. I haven't test your datasets yet. I will have a try to find the issue. Thanks!

[jetrz]

The exact commands I used for each are:
HG002: hifiasm --prt-raw --write-paf --write-ec --dual-scaf --ont -o hg002_ont -t16 -1 /path/to/pat.yak -2 /path/to/mat.yak /path/to/dataset
Fruit Fly: hifiasm --prt-raw --write-paf --write-ec --ont -o fruitfly_ont -t16 /path/to/dataset

Also, can I check is the tomato dataset above haploid or diploid? Would like to test it myself as well. Thanks again!

[chhylp123]

You can double check tomato here: https://genome.cshlp.org/content/early/2024/11/04/gr.279334.124.abstract

[jetrz]

Will give it a look. Thanks!

[chhylp123]

Hi @jetrz, I have fixed this issue. Please note that there is a potential issue with hifiasm that has been observed: it may produce suboptimal results when the input data coverage is extremely high. A temporary solution could be down-sampling to 60x by filtering out shorter reads. We aim to solve this issue as soon as possible in the next release.

[chhylp123]

Hi @jetrz, please have a try with the current release: https://github.com/chhylp123/hifiasm/releases/tag/0.23.0. I tested it on Fruit Fly and now it has no issue even with high coverage.

[jetrz]

Hi @chhylp123 , thank you for the prompt update. I will retest the runs.

Can I check if this problem was present with the --ul scaffolding as well, and if so would this update have affected that?

[chhylp123]

I feel like it won't affect --ul?