how much sequence is needed to cover a mouse transcriptome

[jrss]

Hi all, I am looking for a way to estimate the sequencing space needed to get a reasonable coverage of a transcriptome. Initially I am interested in rat and mouse, but if there is a rule of thumb or a place where I can check for this type of calculations.
thanks a lot for any thoughts or advice

regards

jose

[krobison]

75M tags is what some vendors shoot for on a mammalian transcriptome (for example, on SOLiD4 this works out to 10 samples / slide). I've been meaning to look at the recent papers to try to work it out, though it can be a slog finding the numbers on a library-by-library basis.

[malachig]

In my experience it depends a lot on what you consider 'reasonable' coverage. If you only want to profile gene level expression you don't need much. If you want to have a fair chance of detecting every isoform of every gene, you need at least an order of magnitude more reads. I have analyzed ~100 mouse and human transcriptomes using 'ALEXA-Seq'. When asked this question and forced to give a specific number I usually say that ~100 million paired reads is a good target. In practice it depends on many factors such as the mappability of your reads (which is highly dependent on the error rate at your center). I have also encountered some tissues that had extremely high expression of a small number of genes, and this consumed a lot of read depth. Other tissues did not suffer from this affect as much. You can see the outcome of analyzing some example mouse tissues sequenced to a depth of ~100 to ~250 million paired reads here:

http://www.alexaplatform.org/alexa_seq/Morgen/Summary.htm