Hello Jianguang, On 8/23/12 11:28 AM, Du, Jianguang wrote:
Hi Jen, Thanks for your help. Do you mean that if I want to find novel isoform/splicing, I need to select "No" under "Use Reference Annotation" when I run Cufflink, and then use iGenome version of reference GTF when I run Cuffmerge?
Yes, according to the tool documentation, this is the method.
Based on your information and some protocols found online, my understanding is that: 1) if use iGenome version of reference GTF, I only need to run Cuffmerge with the Cufflink ouputs, because iGenome version reference GTF already contains attributes such as p_id and tss_id. Then the Cuffmerge output can be used for Cuffdiff. Yes, this is the example protocol I shared.
2) however, if I use the reference GTF from Ensembl/UCSC (rather than from iGenome), I need to run Cuffcompare to create p_id and tss_id, which is required for Cuffdiff. This can be tricky, it depends on what order you run the tools with and without the GTF annotation. The protocol in #1 is recommended.
Am I right?
Another question is: should I use iGenome version of reference GTF when I run Tophat if I want to see novel isoforms/splicing? Yes, this is what I intended to answer in my original reply, I apologize if that was not clear. The reference GTF can influence both mapping and assembly. So, both Tophat and Cufflinks. The information on the TopHat web site for the parameter provides more information (see link on TopHat tool form). The tool authors can also be contacted if there are some details that you are curious about that are not covered in the primary documentation: tophat.cufflinks@gmail.com
Others are welcome to add to the thread with their experiences if they have used a reference annotation GTF with Tophat (or chosen not to for a particular reason that they would like to share), Best, Jen Galaxy team
Thanks. Jianguang
________________________________________ From: Jennifer Jackson [jen@bx.psu.edu] Sent: Thursday, August 23, 2012 11:46 AM To: Du, Jianguang Cc: galaxy-user@lists.bx.psu.edu Subject: Re: [galaxy-user] Should I use iGenomes verson of a reference GTF for Tophat?
Hello Jianguang,
When in the analysis process to start using the reference GTF file can depend on whether or not you intend to do any discovery along with differential expression testing. At the TopHat and Cufflinks steps, using reference GTF file can influence how datasets will map and assemble. In general, if your intention is to do discovery (e.g. work with novel isoforms in your data, but not in the reference), then do not add in the reference GTF until the CuffMerge step (to produce the input annotation GTF file for Cuffdiff). But if you want to guide the analysis toward known isoforms, then use the reference GTF.
This is the process our RNA-seq example protocol follows: http://main.g2.bx.psu.edu/u/jeremy/p/galaxy-rna-seq-analysis-exercise
For reference, there are other variations of this on the Cufflinks web site, some that never lead to Cuffdiff, but still may be useful to review. Please see the Cufflinks paper (linked from right side bar as "Protocol" for many more options/discussion. http://cufflinks.cbcb.umd.edu/tutorial.html --> Common uses of the Cufflinks package
The end decision will be up to you, and a few runs with different options may be a useful way to make the final call, but hopefully this provides some resources to help you understand the option,
Jen Galaxy team
On 8/23/12 8:03 AM, Du, Jianguang wrote:
Dear All,
I am analysing RNA-seq datasets for differential splicing events between cell types. These are mouse cells. Jen suggested me to use the iGenomes version of reference GTF to take full advantage of the options in CuffDiff. My question is: should I use this iGenome version reference GTF when I run Tophat?
Thanks.
Jianguang
___________________________________________________________ The Galaxy User list should be used for the discussion of Galaxy analysis and other features on the public server at usegalaxy.org. Please keep all replies on the list by using "reply all" in your mail client. For discussion of local Galaxy instances and the Galaxy source code, please use the Galaxy Development list:
http://lists.bx.psu.edu/listinfo/galaxy-dev
To manage your subscriptions to this and other Galaxy lists, please use the interface at:
-- Jennifer Jackson http://galaxyproject.org
-- Jennifer Jackson http://galaxyproject.org