mRNAseq coverage

[suludana]

Hi,
I would like to know the approximate coverage obtained in a Human mRNAseq PE run (100+100 bp).
I know that it is difficult to estimate the approximate size of the transcriptome, and of course it changes from tissue to tissue and from moment to moment. But I would like to have an idea.
Someone with mRNAseq experience could help me?
Thanks

[SEQme]

The literature is relatively consistent in terms of coverage. The older papers (i.e. Blencowe) suggest that 100 million reads of 36 bp paired end will sufficiently cover a human transcriptome. We have found this to be true with 120 bp paired end runs. 100 million reads will maximally cover the transcriptome with at least 1 read.

[malachig]

As you say. It varies considerably depending on the criteria. I would agree with SAQme's quoted depth as a good target for 'covering the transcriptome'.

However, it also depends on what you are asking of the data (mutation calling, expressed vs. not expressed, detection of minor isoforms, measuring subtle differences in expression between conditions, etc.). If you want to do all of these things for all genes, from the lowest expressed gene to the highest, you may find yourself wishing for more depth.

One way to get a feel for this is to look at the output for many different libraries of different composition.

The following link provides summaries for many libraries that were analyzed in the context of differential and alternative gene expression. These libraries ranged from paired 36-mers to paired 75-mers, and from ~15 million reads per sample to ~500 million, for human and mouse transcriptomes corresponding to ~100 tissues and cell lines.

http://www.alexaplatform.org/alexa_seq/results.htm