<<Problem>>

The cufflinks and cuffdiff results are not consistent with each other. This is killing me.

Does it make sense?

<< Obseration >>  

We have 3 control samples and 3 treated sample. For many genes, their FPKM in cufflinks and cuffdiff are far from consistent. In cufflinks result, for a gene’s FPKM are

control group (sample: 1,2,3)：0， 0， 4.8

treated group(sample: 1,2,3)：0， 0， 6.0

In cuffdiff, the estimated FPKM are

control group： 12.6

treated group：2.0

<<Method>>

Use ucsc gene annotation gtf file, mm9, downloaded from UCSC table database

Use cufflinks on each individual sample.

Cufflinks: galaxy mirror at cistrome, minimal count:10, no quantile normalization, use gtf as reference, no background correction

Use cufflinks on treated groups (3 biological replicates) and control groups (3 biological replicates)

Cuffdiff: galaxy mirror at cistrome, minimal count:10, no normalization, use gtf as reference, no background correction

<<Additional Comments>>

Cufflinks returns 55350 transcripts, while cuffdiff return 55418 transcripts, even though they use the same gene annotation gtf file.

For the 6 cufflinks results (corresponding to 6 samples), the transcript ids are all the same, but the order are not,

<<Question>>

Does it make sense? Or did I do anything wrong?