An odd error with StringTie!!

Question

[seta]

Hi all friends,

I'm running StringTie on the output of STAR. I got bam file from STAR with the following command:

./STAR --runThreadN 8 --genomeDir /home/seta/wrkdir/software/star-2.5.2b/genome --sjdbGTFfile /home/seta/wrkdir/software/star-2.5.2b/genome/anno.gtf --outSAMstrandField intronMotif --outFilterIntronMotifs RemoveNoncanonical --outSAMtype BAM SortedByCoordinate --outFileNamePrefix map1 --readFilesIn /home/seta/wrkdir/Sequensing_files/file1.fastq.gz --readFilesCommand zcat

But when I run the StringTie via the below script:

#! /bin/sh
#$ -S /bin/bash
#$ -N str
#$ -j N
#$ -o str_out
#$ -cwd
#$ -l h_vmem=40G
#$ -pe openmp 4

export OMP_NUM_THEREADS=$NSLOTS
cd /home/seta/wrkdir/software/stringtie-1.3.2d.Linux_x86_64
./stringtie -p 4 -G /home/seta/wrkdir/software/star-2.5.2b/genome/anno.gtf -B -o str1_as/transcripts.gtf -l str1 /home/seta/wrkdir/software/star-2.5.2b/bin/Linux_x86_64/map1Aligned.sortedByCoord.out.bam

But StringTie gave me the error that cannot find the bam file, it's here:

Error: input file /home/seta/wrkdir/software/star-2.5.2b/bin/Linux_x86_64/map1Aligned.sortedByCoord.out.bam
 cannot be found!

But when I run the above stringtie command directly on the terminal (not job schedule), it worked well. Please help me what is wrong with the script?

Thanks

[GenoMax]

This was the problem I was referring to in a previous thread about compilation.

When you compile your own software on a cluster you need to be sure that the location you put it in is accessible from all cluster nodes. At this point you could move the files (that are not accessible based on that error message) to "scratch" space (most clusters will have a common ) and run it from there.

BTW: Keep your data files in a segregated directory. Looks like they are in the program dir now.

[seta]

I have not the problem with STAR, the issue is related to stringtie. However, could you please let me know how I can sure the location is accessible from all cluster nodes?

[GenoMax]

I was focusing on the name STAR in your directory path above hence the error. Post updated accordingly. It is a bit odd that your home directory (based on the name) is not accessible on all cluster nodes.

Answer 1

[Devon Ryan]

Presumably whatever node this is getting run on can't see that directory (or the NFS is really slow). Try an ls on the node first and make sure you save stdout somewhere.

[seta]

Thanks. I have not enough experience with such issues and pass your response to our admin. Just for learning, I searched on the net for doing your advice, but nothing found useful. could you please explain or introduce me some links?

[Devon Ryan]

Your best bet is to bring the cluster admin a coffee on Monday and ask him/her to give you a little overview of how this particular cluster is structured. In some larger clusters, you might find that a certain path is available on some (but not all) nodes. There are typically very good reasons for this, but you'll need an overview so you know what queue you should be submitting things to. Anyway, the admin should understand the error message you're seeing, since obviously if you can see the file on the head node then it's not being seen (at that location at least) on at least some of the cluster nodes.

[seta]

Thanks, Devon. Hope coffee solve the problem!

[Devon Ryan]

If that doesn't do the trick then I recommend cake as well. I kid you not, it can help.