Entering edit mode
9.8 years ago
Wai Yi Leung
▴
60
Dear Kai,
I found a possible bug in one of the functions: GetRealStart4Deletion.
See a fragment from gdb:
There are 0 split-reads mapped by aligner.
report closeMappedReads
search far ends
Far end searching completed for this window.
update FarFragName
update FarFragName done
save interchromsome SR
Searching and reporting variations
Reads already used: 0
Far ends already mapped 17555
Checksum of far ends: 312599851
Searching some type of variant, replace this with the correct name in child class ...
Program received signal SIGSEGV, Segmentation fault.
GetRealStart4Deletion (TheInput=..., RealStart=@0x7fffffffbea0: 16535, RealEnd=@0x7fffffffbea8: 22020693) at pindel.cpp:2058
2058 while (TheInput[PosIndex] == TheInput[End]) {
(gdb)
The "TheInput" in the context is the human chrM with a size of 16569. I suspect the End is larger than the input itself.
I've tested versions:
0.2.4t, 0.2.4w, 0.2.5 (latest checkout from github). All of these version have the same "error".
Hello @liangkaiye. I am getting a segfault using pindel + breakdancer. All other events are output by pindel except for these:
Looking at the coredump:
Segfault happened @ 889
That looks to be the start of the problem.
Have you seen this before? I am currently using (Pindel version 0.2.5a4, May 2 2014.), Is there a specific version of breakdancer I should use too?
Also, I updated pindel to 0.2.5a7 and ran wihout the breakdancer output. It went further than before but I eventually get this error:
Hi liangkaiye,
I also have a segfault in this function, with 0.2.5a7. I don't have it narrowed down to a small sample but can provide a bam with reads aligned against rice which is much smaller than human at least :)
[ edit ]
by this function I meant, I saw
isBreakDancerEventin my initial stack trace. I'm trying to track it further but valgrind is slooowww, probably take a week to show me anythingand I also can complete w/o using breakdancer input, but we see much better results with candidate site lists so...
[/edit]
it generally happens after looking at 10-20 mb of a chromosome, running each chr separately it would crash after last 'Looking' entry in the log.