Hello, I have an issue regarding the step: ALIGN:A...
# raredisease
j
Hello, I have an issue regarding the step: ALIGN:ALIGN_BWA_BWAMEM2_BWAMEME. This step failed on many cases:
[93/255ffe] NFCORE_RAREDISEASE:RAREDISEASE:ALIGN:ALIGN_BWA_BWAMEM2_BWAMEME:BWAMEM2_MEM (SRR8082060_LNUMBER95) [  0%] 0 of 179, retries: 73
I go to the
work/
directory to check some of the error message. So far as I see the error message is always:
samtools sort: truncated file. Aborting
I checked raw fastq files and they are not broken. I also tried to run bwa + samtool in two steps outside of the pipeline using some of the failed samples. I don't see the error and the sorted bam files can be generated successfully. I'm wondering if any of you happen to know how to solve this issue? Many thanks in advance.
r
Do you still have this issue @Jiasheng If you are able to run the processes outside of the workflow, this could be a resources issue. Perhaps the allocated amount of memory if too low?
j
I've solved the issue. I tried to run the process outside of the workflow, but still didn't work. Then, I split the two steps - first
bwa-mem2
then
samtools
, and finally it turned out the problem happened at the first step. There's something wrong with
bwa-mem2
its internal AVX-512 build, which can cause memory corruption or segmentation faults. After switching from the AVX-512 build to the AVX2 build, the alignment runs stabilized and these memory-corruption issues disappeared. I actually cloned the
nf-core/raredisease
pipeline locally and also modified this step (from a streaming pipeline to a two-step process). Since then, I haven’t seen a single error. This entire debugging process took me almost two months, and I hope this workaround can help someone else facing the same issue.
Oh. I should clarify that only one failed case successfully processed outside of the workflow, and I posted the issue when I fortunately see this one case. However, when I tried some other cases, they all failed even outside of the workflow.