MethylDackel: MethylDackel core dump Ubintu 18.04 (problem happend again in issue #95)

First, I have no idea why I cannot re-open my previous issue so I’m writing problem here. (I’m sorry for that)

As you suggested in closed issue #95(https://github.com/dpryan79/MethylDackel/issues/95#issue-619857769) I have tested 0.5.1 but still in the same problem

This is the full backtrace of GDB

Thread 2 "MethylDackel" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffff263f700 (LWP 29417)]
cust_mplp_auto (iter=iter@entry=0x7fffec004e70, _tid=_tid@entry=0x7ffff263ee84, _pos=_pos@entry=0x7ffff263ee88, n_plp=n_plp@entry=0x7ffff263ee94, plp=plp@entry=0x7fffec056280) at pileup.c:288
288	        if (iter->pos[i] == iter->min) {
(gdb) bt full
#0  cust_mplp_auto (iter=iter@entry=0x7fffec004e70, _tid=_tid@entry=0x7ffff263ee84, _pos=_pos@entry=0x7ffff263ee88, n_plp=n_plp@entry=0x7ffff263ee94, plp=plp@entry=0x7fffec056280) at pileup.c:288
        i = 0
        ret = 0
        new_min = 18446744073709551615
#1  0x000055555555f101 in extractCalls (foo=0x7fffffffde60) at extract.c:391
        config = 0x7fffffffde60
        hdr = 0x7fffec003fc0
        iter = 0x7fffec004e70
        ret = <optimized out>
        tid = 0
        pos = 0
        i = <optimized out>
        seqlen = 1000012
        type = <optimized out>
        rv = <optimized out>
        o = <optimized out>
        bedIdx = 0
        n_plp = 0
        strand = <optimized out>
        direction = <optimized out>
        nmethyl = <optimized out>
        nunmethyl = <optimized out>
        nOff = <optimized out>
        nVariant = <optimized out>
        localBin = 0
        localPos = 0
        localEnd = 1000001
        localTid = 0
        localPos2 = 0
        lastPos = 0
        nVariantPositions = 0
        plp = 0x7fffec056280
        seq = 0x7ffff1d4a010 'N' <repeats 200 times>...
        base = 65 'A'
        context = "HG"
        tnc = <optimized out>
        data = 0x7fffec056250
        lastCpG = 0x0
        lastCHG = 0x0
        os = 0x7fffec000b60
        os_CpG = 0x7fffec000b80
        os_CHG = 0x7fffec000ba0
        os_CHH = 0x7fffec000bc0
---Type <return> to continue, or q <return> to quit---
        fai = 0x7fffec001db0
        bai = 0x7fffec0050a0
        fp = 0x7fffec013a70
        __PRETTY_FUNCTION__ = "extractCalls"
#2  0x00007ffff7c29669 in start_thread (arg=<optimized out>) at pthread_create.c:479
        ret = <optimized out>
        pd = <optimized out>
        unwind_buf = {cancel_jmp_buf = {{jmp_buf = {140737260025600, -269718404055552016, 140737488346494, 140737488346495, 140737488346496, 140737260023744, 269688545975873520, 269700372400975856}, 
              mask_was_saved = 0}}, priv = {pad = {0x0, 0x0, 0x0, 0x0}, data = {prev = 0x0, cleanup = 0x0, canceltype = 0}}}
        not_first_call = 0
#3  0x00007ffff7614323 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
No locals.

If you have pre-compiled binary of stable version, can you let me know?

Thanks.

About this issue

Original URL
State: closed
Created 4 years ago
Comments: 30 (11 by maintainers)

Commits related to this issue

Temporarily work around #99 — committed to dpryan79/MethylDackel by dpryan79 3 years ago
Bump MethylDackel back to 0.5.1 for new conda build See https://github.com/dpryan79/MethylDackel/issues/99#issuecomment-781886545 — committed to ewels/nf-core-methylseq by ewels 3 years ago
Bump MethylDackel back to 0.5.1 for new conda build See https://github.com/dpryan79/MethylDackel/issues/99#issuecomment-781886545 — committed to ewels/nf-core-methylseq by ewels 3 years ago
Fix #99 — committed to dpryan79/MethylDackel by dpryan79 3 years ago

Most upvoted comments

Version 0.5.2 contains a reimplementation of the paired-end overlap detection functionality (this is slightly different than what comes with htslib, since it needs to take BS-seq strandedness into account), which was the root cause of the incompatibility with htslib 1.10+. The constructor/destructor functionality pointed out by @jmarshall really helped to clean things up in this regard. Note that 0.5.2 is currently only compatible with htslib 1.11 and not earlier versions, due to https://github.com/samtools/htslib/pull/1127. I hope this issue can now be closed for good 😃

dpryan79 on Feb 28, 2021

I can completely reproduce this with the binary distributed by conda, but oddly enough it doesn’t happen with a regular local build. I’ll make a new conda-built binary locally and track down the cause of this.

dpryan79 on Jun 28, 2020

Looking again, I can see that methyldackel-0.5.1-hee625c5_0.tar.bz2 was used and not the new 0.5.1--hed50d52_1 so I’m pretty sure that this is what the issue was. I’ll try restarting the jobs this evening.

ewels on Feb 19, 2021

@ewels Until I have a proper fix for this supporting htslib 1.10 and 1.11 I’ve made a new build of MethylDackel 0.5.1 pinned to htslib 1.9. That should be available in bioconda in an hour or so, can you try that with the nextflow pipeline?

dpryan79 on Feb 19, 2021