Segmentation fault

[MartinPippel]

Hi,

I have used hifiasm 0.15.3-r339 to assemble a large genome but it crashed with a segfault. We tested it on different machines but the result was the same. Here is the log file:
slurm-17204452.txt

When we did a test assembly with 12x CCS (i.e. 6X haploid) read coverage with hifiasm v0.14.2. the assembly finished without any errors and the assembly stats were surprisingly good.

Any further guidance would be much appreciated.

Best,
Martin

[chhylp123]

Looks like the bin files have already been generated. V0.14.2 and v0.15.3 share the same bin files so could you please run v0.14.2 with the new bin files? I just want to make sure if it is the specific issue of 0.15.3. I personally guess another possibility is no enough memory.

[MartinPippel]

thx for your fast reply. I ran v0.14.2 as well and it crashed at the same point. For the assembly I reserved 5Tb RAM and according to slurm the peak Mem was "only" 1.4Tb. I just submitted it again with 40Tb on the bigmem machine. Is there any option to run hifiasm in debug mode to provide you with some more details?

[chhylp123]

I see. Probably it is a hifiasm issue. It is hard to figure out what happened only with debug information. May I ask what's the size of bin files?

[MartinPippel]

418G ec.bin
190G ovlp.source.bin
55G ovlp.reverse.bin

[chhylp123]

Thanks a lot. I wonder if you can share the bin files with us, so that we can figure out what happened. I'm also pretty curious if it is a bug of hifiasm.

[MartinPippel]

Its confidential data so I have to ask the owner first. I will keep you in the loop.

[MartinPippel]

we got the OK to provide you with the bin files. To whom should I send the download link?

[chhylp123]

Thanks a lot. Could you please send the bin file to hcheng@jimmy.harvard.edu ?

[chhylp123]

Solved by version 0.15.4-r347.