Using Cufflinks to calculate cassette exon inclusion rate

[chaodai]

My question is that can I use cufflinks to calculate cassette exon inclusion rate. Cufflinks output file transcripts.gtf provides isoform expression, is it reasonable to calculate cassette exon expression as the sum of FPKM of the isoforms containing the exon, and gene expression as the sum of FPKM of all the isoforms which belong to the gene, then take the ratio as cassette exon inclusion rate?
I wonder whether this method is accurate enough compared with using junction reads to calculate cassette exon inclusion rate directly.

[yerbol]

Also interested in this question. Does anybody know?

FAQ on Cufflinks site contains such section:
"Can Cufflinks/Cuffdiff give me FPKM for each exon or splicing event?

For the foreseeable future, no."

On the other hand on RNASeq track on ENCODE contains precomputed files -ExonGencV3c.gtf, which are:

"expression level estimates for GENCODE CRCh37.v3c exons in gtf format derived by summing the expression levels in FPKM for all transcripts containing a given exon."

They use Cufflinks too.

And what is the difference in summing FPKMs of transcripts (to obtain total expression level of gene) and summing FPKMs of exons from same transcripts?