We are interested in sequencing a transcriptome. I understand that more depth increases likelihood of identifying subtle differences in gene expression, but if we have no idea the scale of gene expression differences, we are just interested in a first time look at broad patterns, what kind of sequence coverage/depth should we aim for??
I asked a similar question recently; the replies indicated that there is no canonical correct number. I think the basic question is, "given the reported efficiency of the rRNA removal protocol in use by my genome core, and the expected number of mappable reads, how many raw reads do I want to pay for?". Talk with the people at your core (or whoever is running the machines) to see what their experience has been. I got guidance from our collaborator at a major sequencing center, and we are currently targeting 60 million paired end raw reads per sample for mouse transcriptome sequencing using an Illumina HiSeq. For more opinions on this, see a recent paper on RNA-seq in B cells from Vivian Cheung's group (Toung Genome Res. 2011) and some helpful analysis of that paper from Anthony Fejes.
I agree with the comments above but nonetheless thought I'd give you the numbers recommended at an Illumina user group meeting:
3ยด-SAGE: 1-5 M reads
WT-GE: 5-10 M reads
Alt splicing: 10-50 M reads
Novel Transcripts: >50 M reads
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Two quick questions: - Are you doing de novo transcriptome assembly or do you have a reference genome? - How much rRNA contamination do you expect?
we have a reference genome. i'm not sure how much rRNA contamination we will have because we have yet to acquire our samples.