fastq_to_fasta segmentation fault (core dumped)

Question

[Mitra]

Hi, I am trying to convert my fastq files which have been stitched using multiple_join_paired_ends.py under qiime-1.9.1 Now my fastq joined files look okay, but when I try to convert them to fasta I am having trouble. for a single file test I get the error

less fastq2fasta.single.file.sh.e378704

/var/spool/sge_prod/hb01s2b15/job_scripts/378704: line 7: 26760 Segmentation fault      (core dumped) fastq_to_fasta -i RIFSAMPLEA_S1.join.fastq -o RIFSAMPLEA_S1.join.fasta
fastq2fasta.single.file.sh.e378704 (END)

This is the job script that I used.

less fastq2fasta.single.file.sh

#$ -l h_rt=24:00:00
#$ -cwd
#$ -V
#$ -l h_vmem=8G


 fastq_to_fasta -i RIFSAMPLEA_S1.join.fastq -o RIFSAMPLEA_S1.join.fasta


fastq2fasta.single.file.sh (END)

I have looked at the log for the job so I can see that it fails as it is setting up and probably fails at the very beginning of that. Also the log indicates that no memory has been used. I heard, that a segmentation fault is often a problem with allocating memory but it normally occurs late on in the execution of a program when nearly all of the memory has been used and the programm asks for more memory than there is left. But this is definitely not a case here.

Can anybody please help me how to fix this problem? Thanks, Mitra

[GenoMax]

Note: I removed your personal email address from the script above.

I am not sure why this particular program is seg faulting. Has it worked before without any issues? You can't correlate time taken to segfault with amount of memory you are allocating. On fast clusters 8G could be used up in seconds.

As an alternate solution you could use reformat.sh from BBMap Suite to do this conversion by doing reformat.sh in=RIFSAMPLEA_S1.join.fastq out=RIFSAMPLEA_S1.join.fa

[Mitra]

I will try that. Thank you.

And yes obviously fastq_to_fasta works perfectly in all previous occasions. But now recently when I am trying to submit is as a job, it is giving trouble.

And also thanks for deleting my email. I missed that. Mitra

Answer 1

[Ron]

Segmentation fault (core dumped) are mostly related to memory issues. You need to increase your memory. If you are working on a cluster.You can increase the number of slots (> 10) each with atleast 16G of memory. If this does not work,provide more memory or slots.