How to estimate the transcription level?

[SolexaFan]

If I assemble the RNA-Seq data de novo,how can I estimate the transcription level?
My first thought is to calculate the reads frequences of 3' ends,although it seems arbitrary.I use Velvet,generated .afg file may contain useful infomation,but how to analyze this file?Is there any good suggestions?

[swbarnes2]

If you use an aligner, like SOAP or Bowtie, on your de novo assembled transcripts, you can count how many reads hit each RNA by looking at the output, and counting how many times each reference turns up. I don't see why this wouldn't be a fair estiamte of sequence abundance, at least of sequence abundance in the DNA you put on the Solexa. How much things change between your raw RNA and the DNA that goes on the Solexa is another question.

Non-unique hits are a little trickier, but you can set either program to either return every single hit found, or to return a random one each time, so I think you could compensate for either multiple splice variants, or highly homologous regions between transcripts.

I should think at some point you'd want to look at something like a Maq pile-up, to see if you have missed any splice variants, and darned if I can't get velvet to output anything like that, or anything that I can bend into such an display.

[SolexaFan]

Hi swbarnes2,
Non-unique hits and alternative splicing are what I am worrying about.I will try Maq to see if I could deal with Non-unique hits and alternative splicing .Thank you!