Running VG deconstruct failed in the gfa generated from PGGB pipline

[isaamael]

1. What were you trying to do?

Use vg deconstruct to generate vcf from the gfa produced by the PGGB pipeline.

2. What did you want to happen?
The PGGB runs by chromosome, and each chromosome should produce a vcf file

3. What actually happened?
Some chromosomes run successfully and produce results, while some chromosomes fail and report errors.

4. If you got a line like Stack trace path: /somewhere/on/your/computer/stacktrace.txt, please copy-paste the contents of that file here:

ERROR: Signal 6 occurred. VG has crashed. Visit https://github.com/vgteam/vg/issues/new/choose to report a bug.
Please include this entire error log in your bug report!
━━━━━━━━━━━━━━━━━━━━
#1    Object "/public/home/xuruiqiang/software/vg/vg", at 0x20757d5, in raise
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
#3    Object "/public/home/xuruiqiang/software/vg/vg", at 0x5e5053, in __assert_fail_base.cold
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
#2    Object "/public/home/xuruiqiang/software/vg/vg", at 0x5e512b, in abort
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
##3    Object "/public/home/xuruiqiang/software/vg/vg", at 0x5e5053, in __assert_fail_base.cold
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"

━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
#1    Object "/public/home/xuruiqiang/software/vg/vg", at 0x20757d5, in raise
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
#1    Object "/public/home/xuruiqiang/software/vg/vg", at 0x20757d5, in raise
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
1    Object "/public/home/xuruiqiang/software/vg/vg", at 0x20757d5, in raise
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
3    Object "/public/home/xuruiqiang/software/vg/vg", at 0x5e5053, in __assert_fail_base.cold
━━━━━━━━━━━━━━━━━━━━
Crash report for vg v1.50.1 "Monopoli"
Stack trace (most recent call last) in thread 36073:
#24   Object "", at 0xffffffffffffffff, in 
Stack trace (most recent call last)/var/spool/slurm/d/job15597806/slurm_script: line 21: 35163 Segmentation fault      (core dumped) /public/home/xuruiqiang/software/vg/vg deconstruct --all-snarls -P TS --threads 10 $pggbdir/in.fa.gz.e316b50.community.3.fa.out/in.fa.gz.e316b50.community.3.fa.e316b50.417fcdf.7fecd6e.smooth.final.gfa > $outdir/3.vcf

5. What data and command can the vg dev team use to make the problem happen?

6. What does running vg version say?

/public/home/xuruiqiang/software/vg/vg version

vg version v1.50.1 "Monopoli"
Compiled with g++ (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 on Linux
Linked against libstd++ 20230528
Built by anovak@emerald

[adamnovak]

It looks like several threads are failing an assert at once, but the particular assert being failed is not in the log.

I know @glennhickey is working on vg deconstruct currently.

PGGB graphs break some assumptions that we like to rely on for speed, but without the data or the partricular failing assert I can't tell what it is that we can't deal with in this particular graph.

@glennhickey do you know why PGGB graphs might in general not work with deconstruct?

[glennhickey]

deconstruct should work fine on PGGB graphs, though maybe slower due to the reference loop untangling.

But you need the right options. In particular it won't work at all without -e. And I think you need -u to enable the loop entangling.

[isaamael]

Hi ! @glennhickey

I am still not sure what the issue was; however, when I tried again, it worked. Considering that I ran deconstruct on HPC-Slurm without specifying memory, perhaps some tasks required too much memory? I am not sure.

Now, the issue is that I did not specify the --path-traversals and --untangle-travs options during deconstruct as you mentioned. I am not sure what these options do. Will they affect downstream analysis? For example, I need to perform vg call and imputing on the population based on this PGGB graph and vcf.

Appreciation for any suggestions.