JiaoLaboratory
commented
6 months ago This situation is possible, but based on our tests, it occurs almost exclusively in the case of exceptionally long single chromosomes (e.g., when the length of a single chromosome exceeds 300 Mb). Generally, CRAQ consumes more memory when processing high-noise ONT data compared to HiFi data. Even although the memory usage still remains within manageable limits. By the way, could you please provide information on how your 'lgs.sort.bam' was generated? Additionally, if the memory issue cannot be resolved, based on our experience, reducing the usage of long reads (e.g., using only 50X coverage ) is almost no impact on the final results.
yangfangyuan0102
Hi, Dear author The run failed on my machine with 128Gb RAM. My genome is 350 Mb; Pacbio CLR long-read bam is 21 Gb and short-read bam is 11Gb. The memory usage slowly and continuously increased, and finally be killed. Is this normal?
Best wishes
craq -g genome.fa -sms lgs.sort.bam -ngs sgs.sort.bam --sms_coverage 150 --ngs_coverage 65 -t 32 Running CRAQ analysis ......... PARAMETERS: Genome sequence(-g): genome.nextpolish.fa SMS input(-sms): lgs.sort.bam NGS input(-ngs): sgs.sort.bam
Minimum NGS clipped-reads (-sn): 2 Minimum SMS clipped-reads (-ln): 2 Clipping coverRate(-sf): 0.75 Lower clipping rate for heterozygous allele(-hmin): 0.4 Upper clipping rate for heterozygous allele(-hmax): 0.6 Block score benchmarking (-rw): 1000000 Gap[N] is treated with (-gm): 1:CRE Minimum gapsize(-mgs): 10 Break chimera fragments (-b): F Search error region (-ser): T Mapping SMS reads use (-x): map-hifi Mapping quality (-q): 20 Window size for error normalizing (-nw): 35560 Plot CRAQ metrics (-pl): F Alignment thread(-t): 32 Current working at : /mnt/dataset/moth/species/cs/0assembly/9polish/4polish CRAQ output dir(-D): /mnt/dataset/moth/species/cs/0assembly/9polish/4polish/output -------------------------Start Running-------------------------
Running SMS long-reads CRAQ analysis ...... CMD: /mnt/data/miniconda3/envs/craq/bin/../share/CRAQ/src/runLR.sh -g genome.nextpolish.fa -x map-hifi -z seq.size -1 lgs.sort.bam -q 20 -m 2 -f 0.4 -h 0.6 -r 0.75 -a 150 -d 50000 -v F -t 32 worker_pipeline:: Skipping alignment:: [M::worker_pipeline:: Filtering bamfiles] [M::worker_pipeline:: Compute effective SMS coverage] [M::worker_pipeline:: Extract SMS clipping signal] [M::worker_pipeline:: Collect potential CSE|H] [M::worker_pipeline:: Collect potential CRE|H]
Out of Memory and the running was killed here.