Differential expression analysis workflow in Cufflinks

anna_vt · 04-26-2010, 03:16 AM

Hi,

I'm hoping that someone can help me, as I couldn't work out how to do
this from the manual. Would someone be able to give me the steps in a
differential expression analysis?

I have run tophat with the following command for each of my two solexa
sequence.txt RNA-seq files seperately:

tophat --solexa1.3-quals -p 2 -o 101/100315/tophat/
~/software/bowtie-0.12.2/indexes/m_musculus_ncbi37
101/100315/s_2_sequence.txt &> 101/100315/tophat/tophat.out &

I would like to get the expression levels for all Ensembl transcripts.
I have downloaded this gtf file from Ensembl,
ftp://ftp.ensembl.org/pub/current_gtf/mus_musculus,

However when I run
cuffdiff ~/data/gtf/Mus_musculus.NCBIM37.57.gtf
101/100315/tophat/accepted_hits.sam 95/100315/tophat/accepted_hits.sam
&> cuffdiff.out &

or
cufflinks -G ~/data/gtf/Mus_musculus.NCBIM37.57.gtf 101/100315/tophat/accepted_hits.sam &>
101/100315/tophat/cufflinks.out &

I get the following error

Error: duplicate GFF ID 'ENSMUST00000127664' (or exons too far apart)!

I'm pretty sure I've misunderstood the workflow, if someone could give me
an overview of the steps and what gtf file I should be using that would
be great.

Many Thanks
Anna

(Cross posted to Bowtie forum)

Boel · 04-27-2010, 06:26 AM

I get the same error message, and if you look at the transcript ENSMUST00000127664 it is indeed very long and has an intron of size ~ 4.4 Mb. This is way above the default maximum intron length (300,000) and that is why you get this error.

Also, make sure that your GTF file only contains rows for exons, not CDS or transcripts as well. Otherwise all your records in the GTF are duplicated. Maybe you have done this already, but just in case.

Wei-HD · 05-17-2010, 02:11 AM

Hi All,

I met the same error with transcript ENSMUST00000127664, but if I use the Mus_musculus.NCBIM37.56.gtf and the relative index, I did not meet the error. Is that because some annotation has been updated?

Well, I manually deleted the rows which contain transcript ENSMUST00000127664 in the annotation file, then the problem solved. I will try cuffdif later on. But what if people are interested in this gene/transcript? it might not be a good idea to delete? Hopefully someone can give good explanation!

Thanks!

Gangcai · 12-19-2010, 01:52 AM

Quote:

Originally Posted by Wei-HD

Hi All,

I met the same error with transcript ENSMUST00000127664, but if I use the Mus_musculus.NCBIM37.56.gtf and the relative index, I did not meet the error. Is that because some annotation has been updated?

Well, I manually deleted the rows which contain transcript ENSMUST00000127664 in the annotation file, then the problem solved. I will try cuffdif later on. But what if people are interested in this gene/transcript? it might not be a good idea to delete? Hopefully someone can give good explanation!

Thanks!

I met the same problem. Is this problem solved for you?

Wei-HD · 12-19-2010, 02:04 AM

Hi Gangcai,

Sorry I did not figure out a solution about this, I just sticked to the old version index and GTF file, since all my samples were analyzed against the old index (NCBIM37.56). Also I use DESeq R package for all the gene expression level analysis.

I wonder how other SEQers think?

Thanks!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Differential expression, splicing, and promoter use with Cufflinks	Cole Trapnell	Bioinformatics	47	09-24-2015 03:03 PM
Cufflinks 1.0.0: Major new features in assembly and differential expression	Cole Trapnell	Bioinformatics	31	08-26-2011 10:54 AM
Cufflinks differential expression problem	Rachelly	Bioinformatics	2	05-11-2011 11:08 PM
Cufflinks/Cuffdiff significant differential expression	memo	Bioinformatics	5	01-25-2011 09:49 AM
help with differential gene expression with cufflinks and tophat	waterboy	Bioinformatics	1	11-28-2010 09:51 AM

04-26-2010, 03:16 AM	#1
anna_vt Junior Member Location: London Join Date: Jun 2009 Posts: 5	Differential expression analysis workflow in Cufflinks Hi, I'm hoping that someone can help me, as I couldn't work out how to do this from the manual. Would someone be able to give me the steps in a differential expression analysis? I have run tophat with the following command for each of my two solexa sequence.txt RNA-seq files seperately: tophat --solexa1.3-quals -p 2 -o 101/100315/tophat/ ~/software/bowtie-0.12.2/indexes/m_musculus_ncbi37 101/100315/s_2_sequence.txt &> 101/100315/tophat/tophat.out & I would like to get the expression levels for all Ensembl transcripts. I have downloaded this gtf file from Ensembl, ftp://ftp.ensembl.org/pub/current_gtf/mus_musculus, However when I run cuffdiff ~/data/gtf/Mus_musculus.NCBIM37.57.gtf 101/100315/tophat/accepted_hits.sam 95/100315/tophat/accepted_hits.sam &> cuffdiff.out & or cufflinks -G ~/data/gtf/Mus_musculus.NCBIM37.57.gtf 101/100315/tophat/accepted_hits.sam &> 101/100315/tophat/cufflinks.out & I get the following error Error: duplicate GFF ID 'ENSMUST00000127664' (or exons too far apart)! I'm pretty sure I've misunderstood the workflow, if someone could give me an overview of the steps and what gtf file I should be using that would be great. Many Thanks Anna (Cross posted to Bowtie forum)

04-27-2010, 06:26 AM	#2
Boel Member Location: Stockholm, Sweden Join Date: Oct 2009 Posts: 62	I get the same error message, and if you look at the transcript ENSMUST00000127664 it is indeed very long and has an intron of size ~ 4.4 Mb. This is way above the default maximum intron length (300,000) and that is why you get this error. Also, make sure that your GTF file only contains rows for exons, not CDS or transcripts as well. Otherwise all your records in the GTF are duplicated. Maybe you have done this already, but just in case. Last edited by Boel; 04-27-2010 at 06:35 AM. Reason: one more thing!

05-17-2010, 02:11 AM	#3
Wei-HD Member Location: Germany Join Date: Oct 2009 Posts: 59	Hi All, I met the same error with transcript ENSMUST00000127664, but if I use the Mus_musculus.NCBIM37.56.gtf and the relative index, I did not meet the error. Is that because some annotation has been updated? Well, I manually deleted the rows which contain transcript ENSMUST00000127664 in the annotation file, then the problem solved. I will try cuffdif later on. But what if people are interested in this gene/transcript? it might not be a good idea to delete? Hopefully someone can give good explanation! Thanks! Last edited by Wei-HD; 05-17-2010 at 06:40 AM.

12-19-2010, 02:04 AM	#5
Wei-HD Member Location: Germany Join Date: Oct 2009 Posts: 59	Hi Gangcai, Sorry I did not figure out a solution about this, I just sticked to the old version index and GTF file, since all my samples were analyzed against the old index (NCBIM37.56). Also I use DESeq R package for all the gene expression level analysis. I wonder how other SEQers think? Thanks!