Segmentation fault with Mosaik

[bigs]

I'm getting a segmentation fault when trying to sort my assembly by genome position (MosaikSort command). I've tried this step w/ a small set of reads and reference sequence as well as a larger one (Solexa). I'm using a 32-bit linux machine w/ RedHat E5 and 1GB RAM, although the memory does not appear to be the issue.

Anyone encounter this issue w/ Mosaik?

[obenauer]

Having same problem

I don't have a solution, but am having the same problem. So far I've tried an Illumina data set and a Roche data set, and the first few steps of Mosaik run fine (MosaikBuild for the reference, MosaikBuild for the reads, and MosaikAligner), but MosaikSort crashes with an uninformative error: "MosaikSort: free(): invalid pointer: 0x081b0840".

I'm not using paired end reads in my test data, so I thought this would be a simple and trivial use of MosaikSort. I suspect that the problem might be with the output from MosaikAligner, since it has many options to specify. For an E. coli data set using Roche data, here's the MosaikAligner command I used:

MosaikAligner -in reads.dat -ia reference.dat -hs 15 -mm 8 -a single -m unique -act 20 -p 1 -out aligned.dat -rur unaligned.fq

I'll try different options for the hash size (-hs), mismatches (-mm), hashing strategy (-a), alignment candidate threshold (-act), etc., and see if there's a combination that doesn't crash MosaikSort.

[obenauer]

64-bit binary works

The MosaikSort crash only happens with the 32-bit Linux version. I have a 64-bit machine available, and made it through all the steps, including MosaikSort, resulting in an output assembly and ACE file.

Mosaik is only distributed as binaries, so individual users can't recompile the 32-bit version to fix it. We'll have to wait for the Marth lab to supply it. In the meantime, the 64-bit Linux version worked fine, and I also tried the Windows version, which works too.

[shahid.manzoor]

hi I want to run Mosaik so I install ubuntu OS and download Mosaik but the problem is when I try to execute MosaikBuild it give error that Binary file can't be executed. Any body can help me about this error.

[obenauer]

Maybe wrong binary?

hi I want to run Mosaik so I install ubuntu OS and download Mosaik but the problem is when I try to execute MosaikBuild it give error that Binary file can't be executed. Any body can help me about this error.

I think you might get the "Binary file can't be executed" error if you run the 64-bit version of Mosaik in 32-bit Linux, or vice versa. Which version of Ubuntu did you install, 32-bit or 64-bit? The most common Ubuntu downloads are the two ISOs below, where the amd64 version is the 64-bit one.

ubuntu-9.04-desktop-amd64.iso
ubuntu-9.04-desktop-i386.iso

Likewise, you'll need the matching version of Mosaik:

Mosaik-0.9.891-Linux-x64.tar.gz
Mosaik-0.9.891-Linux-x86.tar.gz

As I wrote above, I'd recommend using the 64-bit versions of both in Linux, since the 32-bit version of MosaikSort crashes.

[SMCG]

MosaikJump error std:: bad alloc

I have tried my data on both the 32 and 64 LINUX releases, and I get this error. I have tried the Windows release of the program on my own computer and it works (with the same data), but my laptop will not be sufficient to run a full analysis. I am getting the error std:: bad alloc error on both my data and the C. elegans test data that come with the program. However, the RunAnalysis package that comes with the Unified Release works fine. Does anyone have any insight?
Thanks in advance for any information you can provide!

[SMCG]

Mosaik Aligner

So, I just posted the MosaikJump allocation error question above, and I just tried to run the alignment without specifying a Jump database and received the following "ERROR: unable to allocate enough memory for the DNA hash map." I running this on a huge cluster, though, and set Mosaik to use 6 processors.....

[pedrolance]

For me none of the executable programs work, in neither of the Linux versions. Without any parameters or with, all the programs in the bin directory give the error:

ATAL: kernel too old
Segmentation fault

Probably is a simple error but I have no idea how to fix it. Any help?

[yosemite]

For me none of the executable programs work, in neither of the Linux versions. Without any parameters or with, all the programs in the bin directory give the error:

ATAL: kernel too old
Segmentation fault

Recompiling the binary executable programs from source code helps.

[snownebula]

Hi guys,

The "FATAL: kernel too old" error indicates that your current Linux kernel is older than the one used when compiling the binaries.

The runtime C library, glibc, is compiled to be tied to a certain kernel version.

For example, the current binaries were compiled on a machine with a kernel in the 2.6.x series. Those binaries will work fine on any machine using a kernel in the same 2.6.x series, but will bomb if you have an older 2.4.x kernel.

Most distributions have been using 2.6.x kernels since mid-2004. Some distributions have a slower uptake of linux kernels. e.g. Red Hat Enterprise Linux introduced the 2.6.x kernels only a few months ago (mid-2009).

If you are unfortunate enough to get the "kernel too old" error, but can't easily upgrade the distribution - you can download the MOSAIK source code and compile it. Just enter the src directory and type "make; make utils". The only requirement there is a relatively modern C++ compiler (gcc 4.2 or above).

The source code can be downloaded here:

http://code.google.com/p/mosaik-aligner/

Let me know how things go.

Cheers,

// Michael

[aporter]

Hi Micheal,

I just downloaded the latest version for Linux x86 and am getting the same error:

FATAL: kernel too old
Segmentation fault

when attempting to run any of the Mosaik apps.

Our kernel is of the 2.6.x series:

Linux xxx 2.6.9-55.0.12.ELsmp #1 SMP Wed Oct 17 08:15:59 EDT 2007 x86_64 x86_64 x86_64 GNU/Linux

This is on a computing cluster, but all the nodes have the same setup. We have gcc version 3.4.6 20060404 (Red Hat 3.4.6-8). Upgrading for the whole cluster is not an option at this point. We may be able to upgrade one of our test machines for compiling, but I am wondering if you have another work around in the meantime?

Thanks much,
Alyssa

[Broadie]

Segmentation fault with MosaikSort

Hello,

I am running into this (or a similar) problem as well, when trying to use MosaikSort. We have version 1.0.1307 installed, and I get a segmentation fault on phase 2 of 2: restitch serialized alignments step. I have confirmed that we are running 64 bit Linux so I don't think that is the problem either. Any help would be appreciated.

[dir]

I think the problem occurs with large files. I can handle 1gb files, but 4 gb files with >4M entries don't.

[kumardeep]

I am struggling with the alignment of my .dat file (4Gb) with the reference build file hg19.dat. Both the dat files have been made using MosaikBuild
(i) MosaikBuild -fr hg19.fa -oa hg19.dat
(ii) MosaikBuild -q sample.fastq -st illumina -out sample.dat #for single -end
so far so good, but in the following command its giving error:

(A)
./bin/MosaikAligner -in sample.dat -out sample_unique.bam -ia hg19.dat -p 24 -annpe MOSAIK-2.2.3-source/networkFile/2.1.26.pe.100.0065.ann -annse MOSAIK-2.2.3-source/networkFile/2.1.26.se.100.005.ann -om


Aligning read library (24653305):
Segmentation fault (core dumped)

Then I tried to use jump command as follows

(B) ./bin/MosaikJump -ia hg19.dat -hs 15 -out hg19_15

it is also giving same error after making " Hashing reference sequence:" step

PS : my fastq is a cell line's RNA -seq file of size ~2 Gb which is generating .dat file of 4.2G
Also, my cluster has 500GB RAM and 3TB space.

[Brian Bushnell]

I suggest you try something other than Mosaik; it's never performed well in my tests.