Cufflinks Segmentation Fault

Question

[User 7998]

Hi,

I keep getting a segmentation fault when using cufflinks and no files are generated. The command I use is

cufflinks -G /data/shared/genomes/Btau_4.0/Bos_taurus.Btau_4.0.62.gtf --frag-len-mean 201 --frag-len-std-dev 81 accepted_hits.bam

and I've tried it with and without the options shown. I didn't use -o option because I assumed the files were generated automatically.

Any advice would be welcome.

Regards,

K.

[Istvan Albert]

as pointed out check that your inputs are all in the right format, even beyond the obivous, no zero lenght intervals or other oddities. One good strategy is to keep halving your datasets and see if the problem persists until you are down to one line...

[brentp]

This will be hard to debug even with more information. can you add the out put of head Bos_taurus.Btau_4.0.62.gtf and samtools view -h accepted_hits.bam | head -n 30

is there any output before the segfault?

[Gww]

Did you compile cufflinks yourself? if so what version of boost are you using?

[David Quigley]

Istvan's comment is very good; I would suggest doing it the other way, though-- create the absolute minimal input set first just to see if your setup is otherwise good. That helps establish immediately whether the problem is systemic or somehow coming from your data. I usually find these things are driven by edge cases in the data.

Answer 1

[Pierre Lindenbaum]

As far as I understand cufflinks is a c++ program; you could try to run it using gdb. Something like

$gdb cufflinks
> run -G /data/shared/genomes/Btau_4.0/Bos_taurus.Btau_4.0.62.gtf --frag-len-mean 201 --frag-len-std-dev 81 accepted_hits.bam

but it may requires to re-compile your sources with the '-g' flags and without any optimization ('-O3' or '-O2' flags)

or , if it is available you can try to run your program with truss

Answer 2

[Travis]

I am getting a segmentation fault when I try to use -G with a gff file where the first column has the format "chr1" etc.

If I use a file without the "chr" prefix, the program runs fine. However my results are all zeros as my sam contains chromosome IDs with the chr prefix!

Answer 3

[scchess]

I also got the same problem. While cufflink isn't very informative, you get this mostly likely because there is something weird in the alignment file.

Please take a look at the alignment file. Convert the BAM to SAM and look at the rows. Does the reads mapped to something that you expect? I got the error because I used a wrong file to build the index. All my reads were mapped to strange chromosomes.