Segmentation error (core dumped) in velvetg

[Lakshmi_ISU]

Hi All,
I'm a newbie to the NGS data analysis and still learning through trial and error!
I'm running velvetg in Cygwin on my computer (64GB RAM). I keep getting segmentation fault (Core dumped) error
I have the latest version of velvet and I'm trying to assemble a bamboo genome with illumina data (100bp short reads)

Please can someone help me? Thanks a lot!

[winsettz]

segfault is a pretty vague error. You can try recompiling in debug mode (see the velvet manual for this) and see if you can figure anything out. How's your memory consumption? Does it use up all available memory and then crash?

[Lakshmi_ISU]

Hi winsettz,
Thank you for your reply.

I recompiled in debug mode. I didn't get any additional description of the error except "Segmentation fault (core dumped)". This is the command that I used to recompile:
"make debug BUNDLEDZLIB=1 MAXKMERLENGTH=99 BIGASSEMBLY=1 LONGSEQUENCES=1 OPENMP=1"
It didn't use all the available memory. Used only about 5GB and then crashed.

Then I recompiled again as follows:
"make BUNDLEDZLIB=1 debug"
and it worked NO segfault error! I'm not sure if I'm doing the right thing here?

Thanks!

[winsettz]

Hi winsettz,
Thank you for your reply.

I recompiled in debug mode. I didn't get any additional description of the error except "Segmentation fault (core dumped)". This is the command that I used to recompile:
"make debug BUNDLEDZLIB=1 MAXKMERLENGTH=99 BIGASSEMBLY=1 LONGSEQUENCES=1 OPENMP=1"
It didn't use all the available memory. Used only about 5GB and then crashed.

Then I recompiled again as follows:
"make BUNDLEDZLIB=1 debug"
and it worked NO segfault error! I'm not sure if I'm doing the right thing here?

Thanks!

I would try recompiling without debug, and leave off one flag at a time (bigassembly, longsequences, etc); something is causing your segfault.

[mastal]

What kmer lengths are you using for your assembly?

Velvet uses more memory for longer kmers, and also uses more memory
when it is compiled with the options BIGASSEMBLY and LONGSEQUENCES.

[Lakshmi_ISU]

Thanks for your quick reply!
This is what I got when I recompile with bigassembly:

$ make BIGASSEMBLY=1
rm obj/*.o obj/dbg/*.o
rm: cannot remove `obj/dbg/*.o': No such file or directory
Makefile:50: recipe for target `cleanobj' failed
make: [cleanobj] Error 1 (ignored)
mkdir -p obj
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/tightString.c -o obj/tightString.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/run.c -o obj/run.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/splay.c -o obj/splay.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/splayTable.c -o obj/splayTable.o
src/splayTable.c: In function ‘inputSequenceArrayIntoSplayTableAndArchive’:
src/splayTable.c:1060:14: warning: variable ‘refElem’ set but not used [-Wunused-but-set-variable]
RefInfo refElem;
^
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/graph.c -o obj/graph.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/run2.c -o obj/run2.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/fibHeap.c -o obj/fibHeap.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/fib.c -o obj/fib.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/concatenatedGraph.c -o obj/concatenatedGraph.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/passageMarker.c -o obj/passageMarker.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/graphStats.c -o obj/graphStats.o
src/graphStats.c: In function ‘exportLongNodeSequence’:
src/graphStats.c:953:4: warning: array subscript has type ‘char’ [-Wchar-subscripts]
velvetFprintf(outfile, "%c", tolower(nucleotide));
^
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/correctedGraph.c -o obj/correctedGraph.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/dfib.c -o obj/dfib.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/dfibHeap.c -o obj/dfibHeap.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/recycleBin.c -o obj/recycleBin.o
gcc -Wall -O3 -D MAXKMERLENGTH=31 -D CATEGORIES=2 -D BIGASSEMBLY -c src/readSet.c -o obj/readSet.o
src/readSet.c:38:18: fatal error: zlib.h: No such file or directory
#include <zlib.h>
^
compilation terminated.
Makefile:109: recipe for target `obj/readSet.o' failed
make: *** [obj/readSet.o] Error 1

[Lakshmi_ISU]

Hi mastal,
I was using kmer length=31

[mastal]

Well, it looks like it is crashing with the error
'src/readSet.c:38:18: fatal error: zlib.h: No such file or directory',

which explains why it doesn't crash if you compile it with the option
'BUNDLEDZLIB=1'.

[Lakshmi_ISU]

This time, I recompiled using the cmd:make 'BUNDLEDZLIB=1'
So I assume it uses the default maxkmerlength?
Anyway, I didn't get a segfault!

Thanks a lot for helping me with this!!

[mastal]

yes, the default MAXKMERLENGTH is 31.

if you want to try a longer kmer length you will have to recompile.

[Lakshmi_ISU]

Many thanks! This solved my problem