Entering edit mode
6.4 years ago
juan.crescente
▴
110
I'm trying to use bowtie2 for mapping reads agains a large genome (17Gb). I'm getting an error. It seems like an issue in the index / or some files missing:
./bowtie2 -x refseq -1w_paired.fq -2 rv_paired.fq -S outrefseq --no-unal
Could not open index file refseq.rev.1.bt2l
Could not open index file refseq.rev.2.bt2l
Segmentation fault (core dumped)
(ERR): bowtie2-align exited with value 139
I built the index this way:
nohup ./bowtie2-build /Data/refseq.fasta refseq &
These seems to be the output files
refseq.1.bt2l
refseq.2.bt2l
refseq.3.bt2l
refseq.4.bt2l
This is the last output of build
tail nohup
bucket 7: 60%
bucket 7: 70%
bucket 7: 80%
bucket 7: 90%
bucket 7: 100%
Sorting block of length 2029111534 for bucket 7
(Using difference cover)
Sorting block time: 03:54:26
Returning block of 2029111535 for bucket 7
Is there anything I'm missing?
Segmentation fault (core dumped)
is a typical memory error. You should increase memory for that.Can I specify the memory in the job? What about this error? Could not open index file refseq.rev.1.bt2l
Do you use a SuperComputer system? If yes, you can specify memory during qsub submission. If not, you need to assign more memory to this analysis in your personal computer. There may have an error during indexing files because of lack of memory, so bowtie could not open index file reqseq.rev.1.bt2l.
It looks like bowtie-build did not produce all the files for the index, and you are missing the ones mentioned in the error message, so the bowtie-build command did not complete successfully. Since you are working with a large genome, you may need more memory, as already mentioned by Mehmet, segmentation faults often mean the computer ran out of memory.
So the seg fault was when running bowtie2, not build. Maybe build was killed and did not finish. I'll re-run and update results in a few days.
Can you specify how much RAM you are using for the job/have available?
47GB is the RAM in the machine, can I increase the assigned memory?