State of the Art - RNAseq

[jparsons]

This question really boils down to the following: If you were to do an RNA-seq experiment today (let's limit the sequencing to illumina) what protocols/kits would you use?

*For single-cell / low-volume (let's assume a few hundred cells)
*For a typical bulk treatment/control experiment

As a bit of background, I've got a few ideas right now that involve

*Demonstrating biases in polyA-selection
*Quantitatively assessing the single-cell seq process

I am wondering whether the first will be of negligible impact because tagmentation makes polyA selection less common, and that left me wondering about the single-cells. Are there clear frontrunners in these areas of prep?

[nucacidhunter]

This question really boils down to the following: If you were to do an RNA-seq experiment today (let's limit the sequencing to illumina) what protocols/kits would you use?

*For single-cell / low-volume (let's assume a few hundred cells)
*For a typical bulk treatment/control experiment

I would use Clontech SMARTer kits for single-cell and Illumina's kits for bulk RNA. There are lots of published methods and also comparative studies of various kits and methods. By the time that comparative studies are published new methods are out so one may not be able to make a performance based decision.

[thomasblomquist]

I would use Clontech SMARTer kits for single-cell and Illumina's kits for bulk RNA. There are lots of published methods and also comparative studies of various kits and methods. By the time that comparative studies are published new methods are out so one may not be able to make a performance based decision.

I recommend you peruse the findings from the FDA sponsored Sequencing Quality Control Consortium project http://www.nature.com/nbt/journal/v3.../nbt.3025.html .

One of the principle findings is that no platform, or kit for library prep, provides TRUE accuracy. There is always bias. There is a large degree of agreement between platforms for detecting FOLD-change. But even this can be different between approaches as FRAGMENTATION is not always uniform from day-to-day tech-to-tech.

The key is to stick to a protocol, and it is probably best to batch cases/controls and process in parallel for RNA-seq experiments.

Other key pieces of information: 1) Favor biological replicates over technical replicates. And 2) ultra-deep sequencing only gets you so much information when you are stochastically sampling biological noise. Be careful how deep you sequence.

Good luck,

This is a big beast to harness.

-Tom Blomquist