segfault w/2500 reads in fastq

[calkan]

lordfast --search CH17-157L1.fasta --seq 1-1.fastq -t 8
| Reading Input | 0.00 | XXXXXXXXXXXXXXX | 226.15 | XXXXXXXXXXXXXXX 2500 |
[New Thread 0x7fffbfe01700 (LWP 32000)]
[New Thread 0x7fffbf600700 (LWP 32001)]
[New Thread 0x7fffbedff700 (LWP 32002)]
[New Thread 0x7fffbe5fe700 (LWP 32003)]
[New Thread 0x7fffbddfd700 (LWP 32004)]
[New Thread 0x7fffbd5fc700 (LWP 32005)]

Thread 2 "lordfast" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffbfe01700 (LWP 32000)]
0x0000000000408bbb in findTopWins (chunk_overlap=76, seeds=0x11dc260, isRev=0, readIdx=1, id=0) at src/LordFAST.cpp:465
465 if(_pf_refWin_cnt[id][winId].readIdx == readIdx)
(gdb) p readIdx
$1 = 1
(gdb) p _pf_refWin_cnt[id][winId]
Cannot access memory at address 0x7ffff86c0ad0

[haghshenas]

Dear Can,

Thank you for reporting this.
Could you please let me know which commit you are using?
And also what is the length of shortest read in your fastq file?

[calkan]

git show
commit f5b2747
Author: Ehsan Haghshenas ehaghshe@sfu.ca
Date: Wed Dec 13 17:20:38 2017 -0800

adding option to select the chaining algorithm

shortest read length is 76

[haghshenas]

lordFAST does not map reads shorter than a threshold.
Now this threshold can be adjusted using -l/--minReadLen option (2ee15a2).
At the moment, reducing this threshold causes extra memory usage and runtime. I have a plan to fix this later.

Please let me know if this fixes the problem. Thanks.

[haghshenas]

BTW, it should be ok to keep -l higher than shortest read. However, reads shorter than -l will be reported as unmapped.

[haghshenas]

Hi Can,

This issue should be fixed now. Can you confirm?

[calkan]

Hi Ehsan

@ezgiebren will check. Did you also implement split read support yet? It is important for us.

[haghshenas]

Great!
Yes split alignment is implemented. But I have tested it for long reads only.
You are interested in mapping short reads to long reads, right?