Entering edit mode
2.9 years ago
slin023
•
0
Hi, I am currently running BLAST 2.11.0 thru conda at HPC. I just downloaded pre-complied database from NCBI a month ago (using
wget "ftp://ftp.ncbi.nlm.nih.gov/blast/db/nt.??.tar.gz" -P nt/ && \ for file in nt/*.tar.gz; \ do tar xf $file -C nt && rm $file; \ done )
and run it locally. However, I keep receiving these error messages:
/tmp/slurmd/job5370717/slurm_script: line 21: 39561 Segmentation fault blastn -db /scratch/mdegenna/slin023/nucleotide/nt -query /scratch/mdegenna/slin023/smallGenomecontig/half_genome.fa -outfmt "6 qseqid staxids bitscore std" -max_target_seqs 10 -max_hsps 1 -evalue 1e-25 -num_threads 40 -out /scratch/mdegenna/slin023/BTK_TUTORIAL/FILES/ASSEMBLY_NAME5.ncbi.blastn.out
Which might indicate core dump, so I switched to a different partition with higher mem. still the same messages. Here is my script:
any suggestions will be very helpful! thank you for your time!
Can you make sure your index files are intact and present?
You should see something like the following. There should be 51 files in total. Example below is truncated to save space.
Hi, I have tried. All 51 files are in there.
Then next thing is to try and assign more RAM for your process. So add the following in your script and see what happens.
This will assign 30G of RAM to this job. You may need to adjust this up if you still get a seg fault.
Hi, I keep adjusting, switching to highest memory partition, then to 1500G, tried 500G, none of them worked.
Are you able to start a small test job? Use a single sequence and see what happens. Perhaps there is a problem with your
blast+
install.I tested out all possible problems, I found out it was the V5 database corrupted, so I used
update_blastdb.pl
command to fix it, then issue resolved. thank you!