Cuffdiff - ouput problem

[ytmnd85]

I am attempting to use cuffdiff to do a comparison analysis of various yeast transcriptome samples. I have processed the reads via Tophat and compiled transcripts via Cufflinks. I would like to use Cuffdiff to generate a comparison between samples so that I can start compiling which genes are differentially expressed. I have attempted to run Cuffdiff several times, but I keep getting the same "segmentation fault" error. Here is an example of my input:

$ cuffdiff ../ethanol/transcripts.gtf ~/RNA_Seq/Yeast/reads_for_comp/tophat_output/ethanol/accepted_hits_e.sam ~/RNA_Seq/Yeast/reads_for_comp/tophat_output/glucose/accepted_hits_g.sam

I quickly get the following output:

Segmentation fault (core dumped)

Does anyone have any experience with this type of error arising from Cuffdiff, and any suggestions as to how I can correct this?

Thanks in advance for your help.

[blackgore]

I don't have a solution, but I do have the same problem..though not always.

I used tophat and cufflinks with no problems on a total of 7 different samples. running cuffdiff on two particular samples threw the segmentation fault back at me, but worked fine for other samples. Other than the source organism (it worked for mammal, failed for insect), there was no difference in the way that these samples were prepared or analysed up to that point.

[nicolallias]

Hello everyone,
I've found another way (maybe linked with your problem) to get the "Segmentation fault":

cuffdiff  -o ./ input.gtf input1.bam input2.bam

works fine, but

cuffdiff  -o ./another_directory input.gtf input1.bam input2.bam

causes the "Segmentation fault"

The "another_directory is visible and I've the rigth to write in... But I can't reproduce it on any directories (sometimes it works) :-?
I'm using the current version, v0.9.3
Anyone has the same problem ?

[pasta]

Cuffdiff is a great tool but I stopped using it -> too many bugs

[Aurelien Mazurie]

Originally Posted by pasta

Cuffdiff is a great tool but I stopped using it -> too many bugs

Do you know of a good alternative?

[fongchun]

Originally Posted by nicolallias

Hello everyone,
I've found another way (maybe linked with your problem) to get the "Segmentation fault":

cuffdiff  -o ./ input.gtf input1.bam input2.bam

works fine, but

cuffdiff  -o ./another_directory input.gtf input1.bam input2.bam

causes the "Segmentation fault"

The "another_directory is visible and I've the rigth to write in... But I can't reproduce it on any directories (sometimes it works) :-?
I'm using the current version, v0.9.3
Anyone has the same problem ?

This is actually exactly the same problem I was facing until I read your post. It appears that specifying an outdir other than the current directory caused a segmentation fault. So strange.

[flobpf]

Originally Posted by fongchun

This is actually exactly the same problem I was facing until I read your post. It appears that specifying an outdir other than the current directory caused a segmentation fault. So strange.

I had the segmentation fault despite specifying the same outdir as the current directory. However, if you DO NOT specify the output directory, the segmentation fault goes away. Cuffdiff is working for me now. Strange.

[vineeth_s]

Originally Posted by Aurelien Mazurie

Do you know of a good alternative?

Two alternatives exist:

1. Use HTSeq followed by DESeq : you must note though that there is a significant difference between how cufflinks/cuffdiff and HTSeq/DESeq look at mapping for quantification, there is an enlightening discussion here cufflinks/cuffdiff - HTSeq/DESeq debate

2. There is RSEQTools from the Gerstein lab, this I have not used personally, but the sequencing center attached to where I work base their RNA-Seq analysis pipeline on RSEQTools.

Vineeth

[vineeth_s]

Originally Posted by vineeth_s

Two alternatives exist:

there is an enlightening discussion here cufflinks/cuffdiff - HTSeq/DESeq debate

As that thread has quite many extraneous posts, follow the conversation between Lior Pachter (lpachter) and Simon Anders

[fongchun]

Originally Posted by flobpf

I had the segmentation fault despite specifying the same outdir as the current directory. However, if you DO NOT specify the output directory, the segmentation fault goes away. Cuffdiff is working for me now. Strange.

You are right. I specified -o ./cuffdiff_output and it gave me a segmentation fault halfway into the execution. It seems that the only surefire way of not crashing is to NOT specify any output directory.

[poisson200]

Hi All,
If you don't know already, you can submit your cuffdiff problems to tophat.cufflinks@gmail.com

[DZhang]

Hi All,

Besides poisson200's suggestion, please make sure you are using the latest version to see if it helps. I know this package is under active development.

To me personally, different programs generate different numbers (F/R PKM), which must be unified down the road. It is not good for the science.

Douglas
www.contigexpress.com

[selen]

Originally Posted by flobpf

I had the segmentation fault despite specifying the same outdir as the current directory. However, if you DO NOT specify the output directory, the segmentation fault goes away. Cuffdiff is working for me now. Strange.

I'm glad you all figured it out. I had the same problem and now it works with "-o ./" setting..Thanks!!!!