Cufflinks 2.2.0 and segmentation fault

[Mattia Chiesa]

Hi everybody,
Yesterday I properly installed Cufflinks 2.2.0 on my cluster. Then I tried to test it and compare results with previous version (I used the same parameters and previous versions work vey well).
However an error occur after 3 minutes: Cufflinks failed due to "Segmentation fault".
Does anyone know how to solve this problem?
Thanks,
Mattia.

[Steve Federowicz]

I can second the seg fault on Cufflinks 2.2.0.

In particular I am getting a seg fault when I use -C with a contrast sheet in cuffdiff.  If I allow cuffdiff to make all pairwise then there is not a seg fault. It also runs fine with --use-sample-sheet so presumably the file formatting of those sheets are not the issue.

I am also using .cxb files from cuffquant so maybe that is also part of it.

Best,
Steve

[Sataree Khuansuwan]

Hi,

I'm having the same problem. I am new to cufflinks/cuffdiff. I'm also using Cufflinks 2.2.0. Therefore, I have no reference. Have you figured out a solution to this? I am running the program on Mac OS X with 8 GB of RAM. Should this be sufficient?

Thanks,

Sataree

[Sataree Khuansuwan]

Update: I downloaded and ran an earlier version of Cufflinks (2.1.1) with no problem. 

[Mattia Chiesa]

Sataree, Steve,
I thinks it's not a problem related to your system but to this Cufflinks version: I've a cluster with 48 GB per nodes (12 cpu) and I've got the "segfault" error.
I've never have any problem with previuos versions (cufflinks 2.1.1 included).

For the time being, I keep on use 2.1.1 version.

Mattia

[Minglei Wang]

Hi,
I also have got the "segfault" error for many times with v2.2.0. I even got this error with v2.1.1, but the difference is, with v2.1.1, I could re-run the program, then go through it without any problem; with v2.2.0, this error will keep popping out.

[Tonis Org]

Hi, I am also constantly getting the Segmentation fault: 11 error when trying to run the newest verson of Cufflinks v2.2.1. Very annoying :(. I am running Mac OsX v10.9.3 with 12GB of RAM.

[Ian Driver]

Can't do anything with cufflinks without getting Segmentation fault: 11 I've tried switching everything.  The only thing that works is not using any annotation file during tophat or in cufflinks, which is not very useful to me.  Definitely a file formatting issue somewhere, but I can't for the life of me figure out where?

[Ian Driver]

Update: I got around this by using 2.1.1 rather than 2.2.1, which would only give segmentation fault.  But now trying to run cuffquant it only works on about 10% of my accepted_hits.bam files the rest it always gives Segmentation Fault: 11.  No explanation of why some will run and others won't and why all work in cufflinks but not cuffquant.  What is Segmentation Fault:11? Anyone?

[Yu]

I finally got around the segmentation fault on my Mac OS by downloading the source code and recompiling locally on the Mac. Seems that the downloaded binary has some compatibility issues. Not sure if it's just  a Mac issue or with other OS as well.

[Peter Kim]

Same problem with segmentation faults when using precompiled MacOS binary for 2.2.1. I tried building from source but for me, same problem with segmentation faults.

One thing I noticed was that if I convert the BAM file to a SAM file, cufflinks will process the file without problems. This is inconvenient-- extra step and larger files--, but it is a quick workaround for this problem. The fact that SAMs work, and BAMs don't it makes be think the problem is in cufflinks' interface with samtools or the libbam.a library. This is somewhat curious because samtools presumably uses the same library to do the conversion.

Hope this helps people trying to analyze their data as well as give the developers tracking down this problem.

[Sandra Guauque]

I also have the segmentation fault problem with cufflinks 2.2.1 even if I use SAM files. Did someone find a solution for this problem? I need to use the latest version of cufflinks and compare the results with version 2.1.1. Thank you

[Sandra Guauque]

The problem was solved reinstalling the program. The problem may be due to the due version of the boost libraries used first.

[Mattia Chiesa]

I try to reinstall 2.2.1 version with boost 1.47. The installation is ok but I have again "Segmentation Fault" problem.
Have you updated the boost library as well?

[H Liang]

I was using v2.2.0. Cuffdiff gave Segmentation Fault when I specified contrast file with -C option. But it ran without any problem when removing -C contrast.txt.

I then update to v2.2.1. The new version cuffdiff works with or without -C option.

Not sure my boost version. (how to get the boost version info?)

[Xin Huang]

I also got the segmentation fault 11 error when I used the Cufflinks 2.2.1, but when I re-copied the accepted_hits.bam file from the cluster to the local computer, the problem went away. So it might be because of a file transfer issue, just thought I would bring this up.

Good luck,

Xin

[Asura in]

I also got the same error with Cufflinks 2.2.1, and the problem persisted even if I copied the accepted_hits.bam file from my server to local computer.

  % cufflinks -p 12 -o /output_dir -G Mus_musculus.GRCm38.72.gtf ./accepted_hits.bam
  You are using Cufflinks v2.2.1, which is the most recent release.
  [16:08:45] Loading reference annotation.
  [16:09:04] Inspecting reads and determining fragment length distribution.
  zsh: segmentation fault  cufflinks -p 12 -o /output_dir

But problem solved when I switched to Cufflinks v2.1.1. 

[Chirag Parsania]

Hi,

I installed cufflink on MAC. I tried to run it for 3 samples using 1 sh script. For sample 1 it has completed successfully also created the  out files. But for next sample it has thrown an error "Segmentation Fault". After that I tried to use same command for which it has generated output though it thrown an seg fault error. Couldn't find any reason for that

Kindly help

Chirag

[Michele Mehaffey]

The workaround I found, albeit not perfect, is to remove the gene annotations from the gff file.

[David Leader]

I have found two things that can cause this error message when running cuffdiff 2.2.1. No doubt there are more.

1. If the output file already exists (e.g. one started cuffdiff but it failed because of a type in one of the filenames). You need to delete the output file before running it again.

2. If you run Tophat and one of the replicates gives the following warning:

WARNING: read pairing issues detected (check pre_reads log) !

Apparently it is possible to get the sample to run by changing the bam files to sam files, but I just threw out this replicate (and the output file from the dud run) and cuffdiff ran quite happily.

I hope this may help somebody.

David

[David Leader]

In addition to the previous culprits I can confirm v. 2.2.1 as another possibility. I have just been re-running samples on 2.2.1 that previously ran OK on 2.1.1 and got segmentation errors (although Tophat had run fine). I went back to 2.1.1 and the same samples ran fine.

This is paired-end reads on a iMac with 4 cores running El Capitan.

[Ted Toal]

Same problem, seg fault 11 on OSX running CuffLinks 2.2.1, with no options, just the .bed input file.  It crashed after the message:

Inspecting reads and determining fragment length distribution.

I installed version 2.1.1 and it is currently running fine, it got passed the above point.  I checked to see if any options had changed, as reported by the --help output, and the only difference is that poisson normalization has been removed.