Menu

#297 Overlapper Segfault

overlapper
closed-fixed
nobody
None
5
2015-04-03
2015-02-17
Jeremy
No

My run failed after correction during the assembly overlapping step of PBcR. Here is the full content of results/sample/0-overlaptrim-overlap/000001.out. 000002 and 000003 failed similarly. Do you have any recommendations?

STRING_NUM_BITS 31
OFFSET_BITS 31
STRING_NUM_MASK 2147483647
OFFSET_MASK 2147483647
MAX_STRING_NUM 2147483647

Hash_Mask_Bits 25
Max_Hash_Strings 11273
Max_Hash_Data_Len 20002740
Max_Hash_Load 0.750000
Kmer Length 22
Min Overlap Length 80
MAX_ERRORS 1967
ERRORS_FOR_FREE 1

Num_PThreads 4
Max_Reads_Per_Batch 11273
Max_Reads_Per_Thread 704

HASH_TABLE_SIZE 33554432
sizeof(Hash_Bucket_t) 216
hash table size: 6912 MB

check 128 MB
info 0 MB
start 0 MB

Initialize_Work_Area: MAX_ERRORS=1967 allocated 1MB
Initialize_Work_Area: MAX_ERRORS=1967 allocated 1MB
Initialize_Work_Area: MAX_ERRORS=1967 allocated 1MB
Initialize_Work_Area: MAX_ERRORS=1967 allocated 1MB
Build_Hash_Index from 1 to 11273
HASH LOADING STOPPED: strings 11273 out of 11273 max.
HASH LOADING STOPPED: length 20002740 out of 20002740 max.
HASH LOADING STOPPED: entries 8164529 out of 528482304 max (load 1.16).

realloc Extra_Ref_Space max_extra_ref_ct = 16350095

String_Ct = 11273 Extra_String_Ct = 0 Extra_String_Subcount = 2978
Read 4521 kmers to mark to skip
Index built.
Starting 1 11273
Choose_And_Process_Stream_Segment()-- tid 0
Choose_And_Process_Stream_Segment()-- tid 3
WorkArea 0 allocates space 0 of size 16777216 for array 0 through 4092

Failed with 'Segmentation fault'

Backtrace (mangled):

/wgs-4604/Linux-amd64/bin/overlapInCore(_Z17AS_UTL_catchCrashiP7siginfoPv+0x2a)[0x4160fa]
/lib64/libpthread.so.0[0x3cf460eb10]
/wgs-4604/Linux-amd64/bin/overlapInCore(_Z20Process_String_OlapsPciS_j14Direction_TypeP9Work_Area+0x2ad)[0x410f7d]
/wgs-4604/Linux-amd64/bin/overlapInCore(_Z13Find_OverlapsPciS_j14Direction_TypeP9Work_Area+0x5e1)[0x40d301]
/wgs-4604/Linux-amd64/bin/overlapInCore(_Z16Process_OverlapsP8gkStreamP9Work_Area+0x139)[0x40da29]
/wgs-4604/Linux-amd64/bin/overlapInCore(_Z13OverlapDriverv+0x4af)[0x41288f]
/wgs-4604/Linux-amd64/bin/overlapInCore(main+0xfbf)[0x413b3f]
/lib64/libc.so.6(libc_start_main+0xf4)[0x3cf3e1d994]
/wgs-4604/Linux-amd64/bin/overlapInCore(
gxx_personality_v0+0x101)[0x4069c9]

Backtrace (demangled):

[0] /wgs-4604/Linux-amd64/bin/overlapInCore::AS_UTL_catchCrash(int, siginfo, void) + 0x2a [0x4160fa]
[1] /lib64/libpthread.so.0 [0x3cf460eb10]
[2] /wgs-4604/Linux-amd64/bin/overlapInCore::Process_String_Olaps(char, int, char, unsigned int, Direction_Type, Work_Area) + 0x2ad [0x410f7d]
[3] /wgs-4604/Linux-amd64/bin/overlapInCore::Find_Overlaps(char
, int, char, unsigned int, Direction_Type, Work_Area) + 0x5e1 [0x40d301]
[4] /wgs-4604/Linux-amd64/bin/overlapInCore::Process_Overlaps(gkStream, Work_Area) + 0x139 [0x40da29]
[5] /wgs-4604/Linux-amd64/bin/overlapInCore::OverlapDriver() + 0x4af [0x41288f]
[6] /wgs-4604/Linux-amd64/bin/overlapInCore::(null) + 0xfbf [0x413b3f]
[7] /lib64/libc.so.6::(null) + 0xf4 [0x3cf3e1d994]
[8] /wgs-4604/Linux-amd64/bin/overlapInCore::(null) + 0x101 [0x4069c9]

GDB:

/results/sample/0-overlaptrim-overlap/overlap.sh: line 60: 8523 Segmentation fault $bin/overlapInCore -G --hashbits 25 --hashload 0.75 -t 4 $opt -k 22 -k /results/sample/0-mercounts/asm.nmers.obt.fasta -o /results/sample/0-overlaptrim-overlap/$bat/$job.ovb.WORKING.gz /results/sample/asm.gkpStore

Discussion

  • Sergey Koren

    Sergey Koren - 2015-02-17

    Hi,

    Are you using correction with Illumina data? The CA built-in overlapper is not recommended anymore (it only gets used as a fallback). I would recommend getting SMRTportal (blasr and sawriter) installed which will be used by default for Illumina correction if available. If you are using only PacBio data for self-correction, it should be using MHAP instead and not the CA overlapper.

    Sergey

    [Bri removed quoted email]

     

    Last edit: Brian Walenz 2015-02-25
  • Jeremy

    Jeremy - 2015-02-17

    I am not using Illumina data, just Pacbio for self-correction - attempting
    to use MHAP. Is there something wrong with my command?

    /wgs-4604/Linux-amd64/bin/PBcR -pbCNS -partitions 400 -fastq $fastq -s
    ../../pacbio.spec -length 1000 -libraryname $sample -threads $nproc
    -genomeSize 4600000 localStaging=/tmp

    pacbio.spec:

    original asm settings

    utgErrorRate = 0.25
    utgErrorLimit = 4.5
    cnsErrorRate = 0.25
    cgwErrorRate = 0.25
    ovlErrorRate = 0.25

    merSize=14

    grid info

    useGrid = 0
    scriptOnGrid = 0
    frgCorrOnGrid = 0
    ovlCorrOnGrid = 0

    ovlMemory=8GB --hashload 0.7

    ovlHashBits = 25
    ovlThreads = 2
    ovlHashBlockLength = 20000000
    ovlRefBlockSize = 50000000

    for mer overlapper

    merCompression = 1
    merOverlapperSeedBatchSize = 500000
    merOverlapperExtendBatchSize = 250000

    frgCorrThreads = 8
    frgCorrBatchSize = 100000
    ovlCorrBatchSize = 100000

    non-Grid settings, if you set useGrid to 0 above these will be used

    merylMemory = 128000
    merylThreads = 8
    ovlStoreMemory = 8192
    ovlConcurrency = 8
    cnsConcurrency = 8

    merOverlapperThreads = 3
    merOverlapperSeedConcurrency = 3
    merOverlapperExtendConcurrency = 3

    frgCorrConcurrency = 2
    ovlCorrConcurrency = 4
    cnsConcurrency = 4

    Thanks,
    Jeremy

    [Bri removed quoted email]

     

    Last edit: Brian Walenz 2015-02-25
    • Sergey Koren

      Sergey Koren - 2015-02-17

      Your command and spec should be fine though it is quite old and most of the options aren’t necessary (all those starting with mer), as well as the error rates. I would suggest just using the MHAP one provided with our publication (or the one that comes with the E. coli sample dataset).
      http://www.cbcb.umd.edu/software/PBcR/mhap/asm/pacbio.spec http://www.cbcb.umd.edu/software/PBcR/mhap/asm/pacbio.spec

      However, your spec file shouldn’t affect running MHAP. Can you do an ls on Linux-amd64/lib/java to make sure the mhap jar is present. Can you also provide any outputs from the pipeline as it runs?

      Thanks

      [Bri removed quoted email]

       

      Last edit: Brian Walenz 2015-02-25
      • Jeremy

        Jeremy - 2015-02-18

        Sergey,

        I used your updated spec file, the mhap jar is present and it appears to be using MHAP now, instead of blasr. However, it's still failing during the overlapper step of the assembly, after self-correction. The output from PBcR is attached.

        Thanks,
        Jeremy

         
  • Jeremy

    Jeremy - 2015-02-24

    It looks like this segfault was fixed in revision 4624.

     
    • Sergey Koren

      Sergey Koren - 2015-02-24

      Thanks for the update. Glad to hear your issue is fixed.

      [Bri removed quoted email]

       

      Last edit: Brian Walenz 2015-02-25
  • Sergey Koren

    Sergey Koren - 2015-04-03
    • status: open --> closed-fixed
     

Log in to post a comment.