Generally - yes.

From cufflinks you get a GTF which is an attempt to guess what RNA transcripts are present into a given sample. Sometimes cufflinks is over-sensitive but it is hard to proof and people generally trust it for a lack of better approach :)

This GTF you can use to get gene-based or exon-based counts in the count-based methods.

So - if you happen to find a novel transcript in one sample/group and the count of reads for the genomic coordinates for this sample/group and another one - are significantly different - then you found a novel location with significant differential expression.

You can of course combine GTFs from your favourite annotations and from the cufflinks discovery - using cuffmerge...

Good luck :)

Hey Ms.Bozena,

I don't feel that would cause damage to existing entries. Any how why don't you run the cufflinks for novel transcripts & give a diff. out put name and see the difference yourself.

You may follow this way as given in cufflinks webpage:

-G/--GTF

Tells Cufflinks to use the supplied reference annotation (a GFF file) to estimate isoform expression. It will not assemble novel transcripts, and the program will ignore alignments not structurally compatible with any reference transcript.   (or) 

-g/--GTF-guide

Tells Cufflinks to use the supplied reference annotation (GFF) to guide RABT assembly. Reference transcripts will be tiled with faux-reads to provide additional information in assembly. Output will include all reference transcripts as well as any novel genes and isoforms that are assembled

Thank you both. My worries were also based on using this GTF for peak annotation of ChIP-seq data. I have tried it and compared the outputs with my existing GTF.

While it gives some extra information, several peaks were annotated differently and were assigned to a different gene, which was 10 x the distance than that of the original GTF. It appears that cufflinks has increased the length of one of the genes, that the peak was assigned to in the original GTF. I was just wondering whether it reassigned the TSS in few entries? 

But on the input - you have GTF from RNA-seq, and then apply it to chip-seq, right?

Sometimes cufflinks may report a longer form of the gene... checking on alternative TSS around the end of the gene perhaps could help. Also the UTRs may have the "dropping expression" pattern, that can possibly confuse cufflinks (also depending on the sequencing protocol). Or cufflinks may find an extra far exon - based on the junctions found. Hard to say what is in your particular case - the transcriptome structure needs often to be checked individually. The general rule is : trust the annotation with caution - wether curated or coming from cufflinks. :)