View Single Post
Old 12-19-2012, 01:41 PM   #11
zimmernv
Junior Member
 
Location: Dallas, Texas

Join Date: Dec 2012
Posts: 4
Default

Quote:
Originally Posted by areyes View Post
Hey zimmernv,

Is there any obvious change that you would expect any differential exon usage regulation between the transgenics and the controls? (e.g. the inserted gene being somehow related to the splicing machinery?)

I guess the exon bining is an attempt to test for all possible exon usage regulation according by testing all unique non-overlapping exonic parts from any transcript, but at the end its an optional step. You could input in DEXSeq with "real" exons. My guess is that it won't make a huge difference, but let me know if I am wrong!

Alejandro
Hey Alejandro,

I tried it again with "real" exons, and I got 1 differentially-expressed exon rather than 0. So, as you implied, the difference is not substantial (though that 1 hit is very interesting as far as the biology is concerned).

I was expecting a more pronounced change in exon regulation -- the gene that is inserted is a known splicing factor, so a single event of differential splicing is a bit unexpected. I'm using BitSeq at the moment to see if I can reproduce the results.

Speaking of comparisons, I ran into a thread on SeqAnswers in which S. Anders commented that a golden standard for gene expression and exon expression would be to have 20 replicates of one condition vs. 20 replicates of another condition and then using inferential statistics to find the significant hits. I'm curious about the number 20 -- is there a statistical reason for that value? If it can spare you a long technical explanation, could you send me the titles of the relevant papers?

Thanks a ton,
Vincent
zimmernv is offline   Reply With Quote