Entering edit mode
14 months ago
Prasanna
•
0
I tried to run STAR command for RNAseq but I got the following error.
/home/pshekar/RNAseq/STAR-2.7.11a/source/STAR --runMode genomeGenerate \
> > --genomeDir GRCh38.79.chrom1 \
> > --genomeFastaFiles genome/Homo_sapiens.GRCh38.dna.chromosome.1.fa \
> > --sjdbGTFfile gtf/Homo_sapiens.GRCh38.79.chrom1.gtf \
/home/pshekar/RNAseq/STAR-2.7.11a/source/STAR --runMode genomeGenerate --genomeDir GRCh38.79.chrom1 --genomeFastaFiles genome/Homo_sapiens.GRCh38.dna.chromosome.1.fa --sjdbGTFfile gtf/Homo_sapiens.GRCh38.79.chrom1.gtf --sjdbOverhang 62
> > --sjdbOverhang 62
*****!!!!! WARNING: --genomeSAindexNbases 14 is too large for the genome size=248956422, which may cause seg-fault at the mapping step. Re-run genome generation with recommended --genomeSAindexNbases 12*****
Oct 03 18:27:25 ... starting to sort Suffix Array. This may take a long time...
Oct 03 18:27:26 ... sorting Suffix Array chunks and saving them to disk...
Killed
How much memory does you machine have?
Agreed it's a memory problem, I'm wondering if using the appropriate
genomeSAindexNbasesvalue helps in this case.From the documentation
Using the size provided in the error, the value this person should be using is 13.
It's actually 12 - the creators have said on a forum to take the floor when rounding
log2(GenomeLength)/2-1which isn't well represented in the documentation.https://github.com/alexdobin/STAR/issues/972
This is the code i am using... What should i modify to prevent the error
add the flag
--genomeSAindexNbases 12Why did you add it in the middle instead of at the end? Run this:
Could you answer my question?
I have no idea... I am new to bio informatics.. I am sorry
Ask the system administrator. From your prompt, it looks like you're on a login node. DO NOT RUN high memory/long-running jobs on a login node. Contact your sysadmin about requesting a long running high memory node to run your job.
Can you tell me how to find it
I am using my college's cluster computer
Are you using a compute node or just the login node? If you're using a compute node, how much RAM did you request for allocation?
I am using through online cluster
What does that even mean? What is an "online cluster"?
It still shows error
Do you have write permissions to
genomeDirspecified above. It appears that your process is killed right after it tries to write to the disk location.I didn't write permissions for those... How do I do that
If you execute
touch GRCh38.79.chrom1/testfiledo you get an error? If you do then you don't have write permissions to the directory.This may be a good point to stop and familiarize yourself with
unixcommand line, if you are not aware of unix file permissions. Simply issuing commands (without understanding the logic) is not a good way of doing analysis.Thanks There is no error
Can you please suggest how can I modify the code ... Thanks!
Code itself looks fine. If you are trying to run this on a cluster without going through the job scheduler then it is possible that admins may have set something up to kill jobs that are run outside of the job scheduler. You will need to ask the local support if you are not familiar with how the cluster is setup.
Generally admins frown upon people running analysis jobs on login nodes (which is where you seem to be running this from). Jobs are meant to be run via a job scheduler (SLURM/SGE etc).
Thank you very much
Your code had a backslash instead of a forward slash (
GRCh38.79.chrom1\testfile- I fixed it) - the touch test will not yield useful results with that.