Error on Stack Trace step

[zilov]

Hello, thank you for developoing such a nice tool. I've run LJA on diploid lizard genome with this command lja -t 150 --output-dir ./ --diploid --reads ./raw_reads/hifi/duni.ccs.common.fastq and got error on Stack Trace step. Log in the attachment.

lja_log.txt

[ClemSc]

Hi, Also thank you for the nice tool :)

Aaand, I got the exact same error on a diploid springtail genome. Same command as zilov, I tried with and without the diploid option, for the same results. mdbg.cpp:171: void repeat_resolution::MultiplexDBG::FreezeUnpairedVertices(): Assertionin_edges==out_edges' failed. Child process crashed`

Let me know if I can provide more info.

edit: the pipeline was able to finish successfully for another species, using the diploid option.

[seryrzu]

Hello, thank you for your report and your interest in running LJA! Is it possible to share the data, so I can reproduce this issue? My email is abzikadze@ucsd.edu Thank you!

[seryrzu]

Thank you, @ClemSc, I received the data! We have identified the issue and are working with @AntonBankevich on a permanent fix. In the meanwhile, a temporary fix is just to comment out this assert. Specifically, to comment out line 171 in src/projects/repeat_resolution/mdbg.cpp. After that one can restart LJA with the same command extended by
--restart-from rr. It will ensure that LJA starts from (previously failed) repeat resolution stage without rerunning all previous time consuming steps.

I checked on your data @ClemSc and the temporary fix works for me. Also LJA finished within a few minutes when restarting from repeat resolution stage.

[AntonBankevich]

Hi. I have fixed the issue in branch bug_fix. Unfortunately with this fix you can not use --restart-from rr as Andrey suggested since the real problem occurred earlier than repeat resolution step. But you can use --restart-from phase2 . Please let me know if this solution worked for you.

[zilov]

@AntonBankevich @seryrzu thanks for your help! I've got the assembly with 'bug_fix' branch version. Here are the steps for others who faced the same problem: 1) Go to the LJA folder 2) Change the branch to bug_fix: git checkout bug_fix && git pull 3) (!) Recompile LJA: cmake ./ && make 4) Run LJA with --restart-from phase2 parameter

[LeoVincenzi]

Hi everybody, I used the pipeline proposed in @zilov comment, but it does not succeed. Still, it stops at Assertionin_edges==out_edges' failed. Child process crashed ` during the 'Building graph' phase.
I have also another question about the intermediate data produced: The final_dbg.fasta in k5001 folder is the final assembly (the one that in the end should be reported as assembly.fasta? Is there a difference between them?

[oldzhg]

I tried the 'bug_fix' branch version. Same problem, crash on "Assertion `in_edges==out_edges' failed."