help on Cufflinks output: individual exon have same FPKM

[clariet]

I have run cufflinks for a RNA-seq dataset (SOLiD). As far as I understand, the gtf output file contains the expression level for individual exons. But why all exons from the same gene in my file all have same FPKM value? I thought they should vary a little bit at least. Any suggestions will be appreciated!

an example of gene is listed below:

chr12 Cufflinks transcript 34472253 34500730 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";
chr12 Cufflinks exon 34472253 34472862 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; exon_number "1"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";
chr12 Cufflinks exon 34488469 34488571 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; exon_number "2"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";
chr12 Cufflinks exon 34492266 34492474 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; exon_number "3"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";
chr12 Cufflinks exon 34494455 34494556 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; exon_number "4"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";
chr12 Cufflinks exon 34499873 34499961 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; exon_number "5"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";
chr12 Cufflinks exon 34500504 34500730 1000 + . gene_id "NM_130430"; transcript_id "NM_130430"; exon_number "6"; FPKM "10.5263072455"; frac "1.000000"; conf_lo "9.719728"; conf_hi "11.332886"; cov "21.095279";

[Cole Trapnell]

Cufflinks doesn't actually report exon-level expression. The exon-level GTF attributes are the FPKM for the parent transcript, and are there simply for parsing convenience. Sorry for the confusion, we'll note this in the manual.

[clariet]

Thanks very much for the clarification. I was trying to detect any splicing difference between 2 samples. That's why I need the exon-level expression. Do you have any suggestions on this?

Cufflinks doesn't actually report exon-level expression. The exon-level GTF attributes are the FPKM for the parent transcript, and are there simply for parsing convenience. Sorry for the confusion, we'll note this in the manual.