Segmentation Fault with NR Database (130GB) on WSL

[Raviteja-gajula]

Hi,

I am running Kaiju on WSL (Ubuntu) with the RefSeq NR database (130GB), but I keep getting a segmentation fault.

System Details:
OS: WSL2 (Ubuntu)
RAM: 32GB
CPU: AMD Ryzen 5 (6 cores, 12 threads)
Database: RefSeq NR (130GB)
Command used:
kaiju -t /mnt/d/kaiju_db_refseq_nr_2024-08-13/nodes.dmp
-f /mnt/d/kaiju_db_refseq_nr_2024-08-13/kaiju_db_refseq_nr.fmi
-i /mnt/d/DG250124103535/merged_barcode19.fastq
-o output.txt
-z 4 -e 3
Error message:
Segmentation fault (core dumped)
Steps Taken to Fix:
Checked available memory:
free -h
Result: 23GB free, swap 4.1GB free
Verified database extraction and file permissions (all look fine).
Tried reducing header length in .fmi file but still crashes.
Attempted to use fewer threads (-z 2 instead of -z 4) but no change.
Questions:
Could this be due to WSL memory limitations?
Should I try a static build instead?
Any debugging steps you recommend (e.g., gdb stack trace)?
Any help would be greatly appreciated!

Thanks,
G R