Quantifying trascript levels using RNA-seq

[shazzle]

Hi guys,

I thought of asking this:

1. Can we get sequencing data of a certain mrna from RNA-seq, say from heart and muscle, and use it to quantify and make conclusion about transcript levels in heart and muscle? Is it conclusive, i mean, wouldnt there be noise to take care of and some statistics to normalise the levels of capture on the mrna?

2. If 1. is possible, can we extend the quantification theory to ChIP-seq ie. measuring TF levels in cancer/non-cancer groups?

Thanks!

[spenthil]

I am no way qualified to answer this question, but can point you to two papers that I have been meaning to read that do cover this:

http://www.nature.com/nmeth/journal/v5/n7/full/nmeth.1226.html

http://www.sciencemag.org/cgi/content/abstract/320/5881/1344

Pay particular attention to the idea of "5' bias" and random priming bias - which I think are the biggest problems with current techniques. Of particular note is a comment in the nature paper:

RNA fragmentation before copying would also be expected to greatly reduce 5' bias. This protocol gave better overall uniformity than protocols without RNA fragmentation (Supplementary Fig. 1), although some residual and reproducible nonuniformity clearly persists for randomly primed substrates that was not observed in other kinds of Illumina sequencing substrates handled simultaneously, such as chromatin immunoprecipitation sequencing (ChIPSeq) samples

[malachig]

RNA-Seq can certainly be used to quantify transcripts and compare between tissues. If your RNA is of high quality and your libraries are sequenced to sufficient depth the outcome is likely to be more reliable and less noisy than microarrays. I worked with microarrays for several years and with RNA-Seq for the last two years. Initially RNA-Seq has some growing pains but in my opinion it now beats microarrays hands down. It does not rely on prior knowledge of gene annotations and it has a better signal-to-noise ratio, sensitivity and specificity (if you have enough library depth). Furthermore, in our validations of expression and differential expression estimates from RNA-Seq we found the results to be highly comparable to qPCR.