Computing CL motif score, WINEXTRACT terminated

[algaebrown]

Hi! Sabrina
I ran into this problem when running this:

genome=/home/hsher/genome_fa/GRCh38.p13.genome.fa
motifxml=/home/hsher/pureclip_motifs/motifs.xml
motiftxt=/home/hsher/pureclip_motifs/motifs.txt
odir=/home/hsher/pure_output/
ip=/projects/ps-yeolab3/encode/analysis/encode_master/271_01_HNRNPC.merged.r2.bam
fname=${ip##*/}
prefix=${fname%.*}


compute_CLmotif_scores.sh $genome $ip $motifxml $motiftxt $outdir$prefix.clmotif.bed

I tried1 node and 2 processor, '10 processors and 6 processors, both failed

This is the error message:

terminate called after throwing an instance of 'std::bad_alloc'                                               
what():  std::bad_alloc                                                                                   
/home/hsher/miniconda3/envs/pureclip/bin/compute_CLmotif_scores.sh: line 94:  6781 Aborted                 "$WINEXTRACT" -g "$REF" -c "$TEMP_DIR/alignments.bam.regions2.bed" -o "$TEMP_DIR/alignment.covered_regions.sequences.fasta" -uow

This is the output:

Using temporary directory '/scratch/hsher/22199369[0].tscc-mgr7.local/tmp.6vzDCFOJ1s'                       
Load reference file ...                                                                                     
Load candidate sites from BED file ...                                                                      
Extract sequence for windows around candidate sites ...

I am really new to this and doesn't know how to solve this problem.
it seems to be an memory issue. How much memory should I use?
Thanks!

[skrakau]

Hi @algaebrown,

yes, it sounds like a memory issue. PureCLIP itself requires quite a bit of memory, but this script usually not that much, although I never measured the consumption to be fair and it depends on how many regions are covered in your case.

Could you try with more memory? Let me know if the problem persists, or (if more memory doesn't help) you could send me the input files for the winextract.cpp script where the error occurs, i.e. "$TEMP_DIR/alignments.bam.regions2.bed" and "$REF".

Best,
Sabrina