mosdepth error

[pauline-ng]

Hi Brent,

I'm running smoove with a small reference cohort:

smoove call -x --outdir /home/pauline/cnv_problem_files/1-0041-003/go_debug --name 1-0041-003_recal --fasta /myome/share/references/hs37d5/hs37d5.fa -p 12 --genotype /home/pauline/cnv_problem_files/1-0041-003/smoove_ref_cohort_tmdir_intern/HG01710.bam /home/pauline/cnv_problem_files/1-0041-003/smoove_ref_cohort_tmdir_intern/HG01709.bam /home/pauline/cnv_problem_files/1-0041-003/smoove_ref_cohort_tmdir_intern/HG00733.bam /home/pauline/cnv_problem_files/1-0041-003/smoove_ref_cohort_tmdir_intern/HG01695.bam /home/pauline/cnv_problem_files/1-0041-003/1-0041-003_recal.bam

and getting the error:

[smoove] 2022/03/18 02:18:14 starting with version 0.2.6
[smoove] 2022/03/18 02:18:14 calculating bam stats for 5 bams
[smoove] 2022/03/18 02:18:24 done calculating bam stats
bash: line 5:  3860 Killed                  mosdepth -f /myome/share/references/hs37d5/hs37d5.fa --fast-mode -n --quantize 1001: /home/pauline/smoove-mosdepth-862318781 /home/pauline/cnv_problem_files/1-0041-003/go_debug/HG01709.split.bam
panic: exit status 137


goroutine 81881 [running]:
github.com/brentp/smoove/lumpy.check(...)
        /home/brentp/go/go/src/github.com/brentp/smoove/lumpy/lumpy.go:54
github.com/brentp/smoove/lumpy.remove_sketchy(0xc00023e320, 0x45, 0x3e8, 0x7fff7f3d0baa, 0x28, 0x0, 0x0, 0xc00018d640, 0x4, 0x4, ...)
        /home/brentp/go/go/src/github.com/brentp/smoove/lumpy/depthfilter.go:232 +0x1818
github.com/brentp/smoove/lumpy.remove_sketchy_all.func1(0xc0017c8b40, 0x3e8, 0x7fff7f3d0baa, 0x28, 0x0, 0x0, 0xc00018d640, 0x4, 0x4, 0x1, ...)
        /home/brentp/go/go/src/github.com/brentp/smoove/lumpy/depthfilter.go:399 +0x231
created by github.com/brentp/smoove/lumpy.remove_sketchy_all
        /home/brentp/go/go/src/github.com/brentp/smoove/lumpy/depthfilter.go:397 +0x191

I saw there was an OOM issue on your mosdepth repo, is there a way to specify memory with mosdepth inside smoove so this error doesn't occur?

Other information that may be relevant:
It's 1 particular bam that's causing the problem (1-0041-003). I can run 1-0041-003 individually and with a few other reference samples, but when it's run with all the reference files fails. 1-0041-003 is the sample file -- the others are my reference files and have run successfully with lots of other samples.

Thanks,
Pauline

[brentp]

Hi, the memory shouldn't be an issue unless you have less than 2GB of memory per sample available. But, indeed exit status 137 is out of memory. How much memory is available? Maybe you need to specify, e.g. 15GB total and then run the job.

you could also try updating to the latest version of smoove as that will contain other recent changes (though I don't think any directly related to this).

[pauline-ng]

Hi Brent,

Thanks for the tip. What finally worked for me was to reduce the number of threads for the smoove call. (smoove -p)

Note: Specifyingdocker --memoryor docker --memory --oom-kill-disable didn't work for me.