Segfault on example data (CentOS)

[chartl]

I'm running:

$ cat /proc/version Linux version 2.6.32-573.18.1.el6.x86_64 (mockbuild@c6b8.bsys.dev.centos.org) (gcc version 4.4.7 20120313 (Red Hat 4.4.7-16) (GCC) ) #1 SMP Tue Feb 9 22:46:17 UTC 2016

and from the base directory of the github clone I run:

$ rm -rf examples/Pearson && make clean && make && cd examples/ && mkdir Pearson && ../ccm yeast-s_cerevisiae1.global.RMA.nc-no-na-nh 10359 1535 rm -f ccm.o rmm.o ccm rmm gcc -m64 -c -g ccm.c gcc -m64 -c -g rmm.c gcc -m64 ccm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o ccm gcc -m64 rmm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o rmm Reading input file 'yeast-s_cerevisiae1.global.RMA.nc-no-na-nh'... Segmentation fault

[spficklin]

Hi chartl.

Try adding the .txt to the end of the yeast filename. At one point the code did require that the extension be removed but I believe the code was changed to allow it and it looks like the README wasn't updated.

I added a fix so that the code will no longer segfault, it will give you a message about the file not being found. And the README has updated.

[chartl]

Same issue:

$ rm -rf examples/Pearson && make clean && make && cd examples/ && mkdir Pearson && ../ccm yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt 10359 1535 rm -f ccm.o rmm.o ccm rmm gcc -m64 -c -g ccm.c gcc -m64 -c -g rmm.c gcc -m64 ccm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o ccm gcc -m64 rmm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o rmm Reading input file 'yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt'... Segmentation fault (core dumped)

And with full extension ('.txt.gz')

$ rm -rf examples/Pearson && make clean && make && cd examples/ && mkdir Pearson && ../ccm yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt.gz 10359 1535 rm -f ccm.o rmm.o ccm rmm gcc -m64 -c -g ccm.c gcc -m64 -c -g rmm.c gcc -m64 ccm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o ccm gcc -m64 rmm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o rmm Reading input file 'yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt.gz'... Error: EOF reached early. Exiting.

Update:

Unzipping did the trick

$ rm -rf examples/Pearson && make clean && make && cd examples/ && mkdir Pearson && ../ccm yeast.test.txt 10359 1535

rm -f ccm.o rmm.o ccm rmm gcc -m64 -c -g ccm.c gcc -m64 -c -g rmm.c gcc -m64 ccm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o ccm gcc -m64 rmm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o rmm Reading input file 'yeast.test.txt'... Calculating correlations... Writing file 1 of 2: ./Pearson/yeast.test.pc0.bin...

[spficklin]

Oh, you'll have to uncompress the file first.

gunzip yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt.gz

Have you pull the code since my last reply? I t should have warned you that that file couldn't be opened when you tried it with just the .txt extension.

[spficklin]

HI chartl. Did this resolve the issue for you? If so I'll close the issue. If not let me know and we can keep working on it.

[jmwhitha]

We ran into the same issue. This produced a .pc0.bin and .pc1.bin, but then when we tried to run rmm -i yeast-s_cerevisiae1.global.RMA.nc-no-na-nh we had a segfault. We also had a segfault when we tried putting the .pc0.bin extension at the end of the file. It also segfaults with relative and full path.

[jmwhitha]

Sorry, I'm from another group. We are using Ubuntu 15.10.

[jmwhitha]

What OS did you test RMTGeneNet on?

[spficklin]

I'm using Ubuntu 14.04. If you are using the example yeast file you can create the network in this way:

1) Change directories to the 'examples' directory 2) Run the ccm program: ../ccm yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt 10359 1535 3) Run the rmm program: ../rmm -i yeast-s_cerevisiae1.global.RMA.nc-no-na-nh -b 0.930000 4) Generate the network with the perl script: perl ../parse_pearson_bin.pl -b Pearson -t 0.831100 -p probest_ordo yeast-s_cerevisiae1.global

I updated the examples/README.txt to fix some typos that hopefully will help. I just ran the commands above and all worked well.

[jmwhitha]

I tried that, and the first command didn't work. So, I thought it might be the way I edited the Makefile since I had a separate RMTGene folder. So I downloaded the RMTGeneNet-master and used the following Make File. While there was no yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt in the example folder this time, I just copied it from the old RMTGeneNet folder I kept in a separate directory. The command still didn't work even though ccm was now in the folder above the example. I will also post my commands so you can see. I'm thinking now it may have to do with my MKL. Where did you get yours? I have a free trial now, but only for a few more days, and it comes in a very expensive package.

MKLROOT = /opt/intel/mkl

MKLROOT = /opt/intel/composerxe-2011.0.084/mkl

CC = gcc -m64 EXE_DIR = /opt/RMTGeneNet

CCFLAGS = -g

LDFLAGS = -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm

OBJS = ccm.o rmm.o EXE = ccm rmm

all: ${OBJS} ${CC} ccm.o ${LDFLAGS} -o ccm ${CC} rmm.o ${LDFLAGS} -o rmm

ccm.o: ccm.c ${CC} -c ${CCFLAGS} ccm.c

rmm.o: rmm.c rmm.h ${CC} -c ${CCFLAGS} rmm.c

clean: rm -f ${OBJS} ${EXE}

install: all install -m 0755 ccm ${EXE_DIR} install -m 0755 rmm ${EXE_DIR} install -m 0755 parse_pearson_bin.pl ${EXE_DIR}

Here's what it says when I "make" and "make install"

vjw@Thermomics:/opt/RMTGeneNet-master$ sudo make gcc -m64 -c -g ccm.c gcc -m64 -c -g rmm.c gcc -m64 ccm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o ccm gcc -m64 rmm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o rmm vjw@Thermomics:/opt/RMTGeneNet-master$ sudo make install gcc -m64 ccm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o ccm gcc -m64 rmm.o -g -Wall -lm -lgsl -lgslcblas -fopenmp -llapack -lblas -lpthread -lm -o rmm install -m 0755 ccm /opt/RMTGeneNet install -m 0755 rmm /opt/RMTGeneNet install -m 0755 parse_pearson_bin.pl /opt/RMTGeneNet vjw@Thermomics:/opt/RMTGeneNet-master$ ls ccm ccm.o INSTALL.txt Makefile README.txt rmm.c rmm.o ccm.c examples LICENSE.txt parse_pearson_bin.pl rmm rmm.h

[jmwhitha]

vjw@Thermomics:/opt/RMTGeneNet-master$ cd examples vjw@Thermomics:/opt/RMTGeneNet-master/examples$ ls probest_order.txt README.txt samples.txt vjw@Thermomics:/opt/RMTGeneNet-master/examples$ cd README.txt bash: cd: README.txt: Not a directory vjw@Thermomics:/opt/RMTGeneNet-master/examples$ sudo cp ~/Downloads/RMTGeneNet/examples/yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt ./ vjw@Thermomics:/opt/RMTGeneNet-master/examples$ ../ccm yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt 10359 1535 Reading input file 'yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt'... Calculating correlations... Writing file 1 of 2: ./Pearson/yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.pc0.bin... Segmentation fault vjw@Thermomics:/opt/RMTGeneNet-master/examples$

[jmwhitha]

Not sure why but README.txt is an executable in the folder. So is LICENSE.txt, INSTALL.txt, ccm.c, rmm.h, rmm.c, and parse_pearson_bin.pl.

[spficklin]

Very strange. I'm not sure why you're getting a segfault. You have the most up-to-date code? Does it segfault almost immediately after it prints the message about writing file 1 of 2? Or, does it take a bit of time?

I just reran the example again and here's what I get:

../ccm yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt 10359 1535 Reading input file 'yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.txt'... Calculating correlations... Writing file 1 of 2: ./Pearson/yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.pc0.bin... Writing file 2 of 2: ./Pearson/yeast-s_cerevisiae1.global.RMA.nc-no-na-nh.pc1.bin... Done.

[jmwhitha]

Thanks for the response, but I'm still having trouble. As much as I would like to, I may not be able to use this method given the time constraint.

[jmwhitha]

Someone from our help desk helped me get further.

We tried the command below and got the following error. Any idea what's happening? strace ./rmm -i 2016-5-17_RMAnormSummary -b 0.930000

read(5, "\335d\343>\320$~>b\241\323\275$\332\343\275\17[:\276\204'i\276\307\320]=x*\365="..., 4096) = 3416 close(5) = 0 munmap(0x7f882e0e8000, 4096) = 0 mmap(NULL, 1196032, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f882dfa6000 write(1, " calculating eigenvalues for n "..., 62 calculating eigenvalues for n x n matrix of size n = 546... ) = 62 munmap(0x7f882dfa6000, 1196032) = 0 write(1, " testing similarity of NNSD wit"..., 45 testing similarity of NNSD with Poisson... ) = 45 write(2, "gsl: interp.c:38: ERROR: insuffi"..., 78gsl: interp.c:38: ERROR: insufficient number of points for interpolation type ) = 78 write(2, "Default GSL error handler invoke"..., 35Default GSL error handler invoked. ) = 35 rt_sigprocmask(SIG_UNBLOCK, [ABRT], NULL, 8) = 0 write(4, "Threshold\tChi-square\tCut Matrix "..., 37) = 37 tgkill(1816, 1816, SIGABRT) = 0 --- SIGABRT {si_signo=SIGABRT, si_code=SI_TKILL, si_pid=1816, si_uid=0} --- +++ killed by SIGABRT +++ Aborted

[spficklin]

The most recent code changes in the development branch should resolve these issues. I'm marking this as closed. Please reopen if some problem still remains.