Good evening,
I followed the tuxedo protocol as described here. With the "normal" outputs that cufflinks provide (I'm searching on cuffdiff output files, but tell me if I'm wrong), how could I determine which of isoforms detected are actually novel ones?
I'm pretty new on this field, and I've found in many posts that using Cufflinks is the answer to find novel isoforms, or differential splicing too. Now that I followed up the "instructions" I'm stuck on trying to interpret the results, after searching all over I could. Searching on cufflinks manual didn't help me either. :/
Can someone explain me or provide me some tutorials/papers/etc, that might help me on this topic please?
Thanks in advance.
2 answers
Here are two tutorials on RNA-seq, both that involve discussion of Cufflinks that you might find generally useful:
With regard to your specific question. You might find the Cuffcompare and Cuffmerge commands and their documentation useful.
- http://cole-trapnell-lab.github.io/cufflinks/cuffmerge/index.html
- http://cole-trapnell-lab.github.io/cufflinks/cuffcompare/index.html
Cuffcompare has this description:
After assembling a transcriptome from one or more samples, you'll probably want to compare your assembly to known transcripts. Even if there is no "reference" transcriptome for the organism you're studying, you may want to compare the transcriptomes assembled from different RNA-Seq libraries. Cuffcompare helps you perform these comparisons and assess the quality of your assembly.
You can run cuffcompare directly but in recent versions of cufflinks, the cuffmerge command will do this for you. You should have '.tracking' output files. Refer to the 'class code' column of this file. The definition of "transfrag class codes" is provided at the bottom of the Cuffcompare manual page. For example, class "j" is described as "Potentially novel isoform (fragment): at least one splice junction is shared with a reference transcript"
•
written
by
Malachi Griffith
hi Your advice was helpful. Thanks a lot
hi get cufflink manual from enter link description here
Log in to answer this question.
-
The right pipline with cufflinks, cuffmerge and cuffdiff
written by seta
-
cuffdiff notest but cufflinks shows fpkm > 10
written by urmi208 •
-
From HISAT2 to cuffdiff?
written by dina.hesham139
-
Cuffmerge potential error
written by moustafa_abohawya •
-
From Cuffdiff Output to CummeRbund analysis/visualisation
written by Wet&DryImmunology
-
Cuffdiff (weird) output
written by rna-seq_researcher •
-
Cuffdiff splicing.diff output
written by zhengyuguobuaa •
-
Summary of alternative splicing from Cufflinks
written by Daniel Standage
-
Cufflinks With Gtf File
written by GouthamAtla
-
Cufflinks & Cuffdiff Novel Isoform Detection
written by andrew.j.skelton73
I have done this by running cufflinks with no annotation, then using bedtools to basically remove all the regions that intersect with known genes... Or if you're interested in antisense transcription, look for transcripts on the opposite strand from known genes. Then apply some filters on length and rpkm and come up with a set of putative novel transcripts. Hopefully followed up by some kind of lab based verification, because I feel like this is all pretty hazy and based on thresholds. This may not be the right way to do it, but it seemed to make sense to me at the time.