running deFuse on Sun Grid Engine cluster

[Lien]

Dear all,

I'm trying to run deFuse on a SGE cluster to detect gene fusions in paired-end RNA-seq samples.
My code:

sge ./defuse.pl -c /defuse/scripts/config.txt -1 sample1.R1.fastq -2 sample1.R2.fastq -o /defuse_out/sample1 -s sge -p 8


deFuse splits in multiple nodes, and some nodes appear to run fine, while others throw an error:

Success for defuse command:
/cm/shared/apps/defuse/current/scripts/alignjob.pl -c /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse/scripts/config3.txt -j reads.split.049 -l /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse_out/GC011094/tmp -o /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse_out/GC011094 -n GC011094 -p /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse_out/GC011094/jobs/reads.split.049
Return codes: 0
Job output:
Running on cnpew17.uz.kuleuven.ac.be
Starting alignjob reads.split.049
Finding concordant alignments to cdna
Finding concordant alignments to dna
Calculating read statistics
Calculating covariance samples
Calculating expression
Trimming and filtering reads
Finding all cdna alignments
Finding all dna alignments
Finding same gene concordant read ids for cdna
Finding same region concordant read ids for dna
Finding alignments to rRNA
Finding rrna anchored concordant read ids
Finding concordant alignments to prefilter fasta /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse/Hs.seq.uniq
Excluding IG rearrangements
Coallating concordant read ids
Creating list of improper reads
Dividing sam output files
Finished Alignments
Cleaning Up

real 103m15.988s
user 97m46.969s
sys 2m31.687s



A failed node:
Failure for defuse command:
/cm/shared/apps/defuse/current/scripts/alignjob.pl -c /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse/scripts/config3.txt -j reads.split.040 -l /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse_out/GC011094/tmp -o /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse_out/GC011094 -n GC011094 -p /uz/data/avalok/symbiosys/gcpi_r_uz_peter_vandenberghe/lspans0/defuse_out/GC011094/jobs/reads.split.040
Reason:
Job command with nonzero return code
Finding concordant alignments to dna
Job output:
Running on cnpew04.uz.kuleuven.ac.be
Starting alignjob reads.split.040
Finding concordant alignments to cdna


Is there a solution for this?

How come it works for some and fails for others, while they are all part of the same command?