Expressed genes vs Expressed transcripts

[lucasmiguel]

I analyzed differential gene expression using Cuffdiff, but the number of genes expressed is smaller than transcripts expressed. Anyone can explain this?

In my mind, the number of transcripts should be bigger.

[dpryan]

You just wrote that the number of transcripts is bigger, which is what you expect. Did you mean that the number of genes expressed is larger than the number of transcripts? If so, how are you defining expressed? I'm guessing that you're setting a threshold and then noticing that fewer transcripts are expressed since a given gene's signal can get diluted over multiple transcripts (possibly causing all of them to fall below threshold even though the gene as a whole is above threshold).

[lucasmiguel]

I defined that one gene is DE if he has FDR <= 0.05 (5%) and the same threshold for transcripts. It's wrong?

And yes, i meant larger than.

Thanks for replying, dpryan.

[dpryan]

Well expressed is very different from DE. It's not surprising to find fewer DE transcripts for the reasons I mentioned in my first post (I don't remember if cuffdiff looks at genes with only 1 transcript in the transcript output, but if not then that'd further decrease things).

[lucasmiguel]

Ok. Thanks dpryan, helped a lot.