[davemessina]

Hi,

Great question.

We're developing tests around a recently discovered type of RNA called circular RNA. Tissue-specific circular RNAs are present in blood, and so that's how it's possible to use a blood test instead of having to go to tissue like you would with traditional mRNA-seq.

[daemonk]

I don't quite understand how just using circ. RNAs would give you better predictions. From the little I've read about circ. RNAs is that they can form from normal RNAs and possibly function in sequestering regulatory small-RNAs.

So you can detect circ. RNA that come from specific tissues and partition those reads out as expression for those tissues? What kind of coverage do you need to do that? Are mRNA circularization rates consistent among different genes?

[jarmstropreamp]

Hi daemonk, Jon Armstrong from Cofactor Genomics here. You hit on some excellent points.

So you can detect circ. RNA that come from specific tissues and partition those reads out as expression for those tissues?

In reality, we are detecting circRNA that arises or predominates during a specific disease state and then partition those reads out as a signal for the disease state. On a related note, the circRNA molecule resists degradation much more than linear RNA molecules, leading to a longer half-life and less variability in signal from tissue, plasma, and exosomes.

What kind of coverage do you need to do that?

Currently, 100's of million of reads need to be generated to just detect the highest expressed circRNAs. We have a patent pending circRNA enrichment technology (supported by a large NIH phase II grant) that increases the ratio of circRNA in the total RNA pool by logs, thus one could sequence logs less reads for the same level of detection. The coverage cutoff for detection can be "tuned" for assay sensitivity.

Are mRNA circularization rates consistent among different genes?

CircRNAs do not seem to be translated into proteins and probably perform multiple regulatory functions. Also, the level of a linear form does not correlate with the level of the circular form for the same gene. As with linear mRNA production, which is not consistent from gene to gene or tissue to tissue, we observe the same phenomenon with circRNA.

Hope this info helps.

[niekmaas]

Finally a topic on hackernews that I am getting my PhD in..! Anyway, do you have any data indicating what percentage of circRNA is "free" in the plasma versus inside of other entities like platelets or extracellular vesicles (exosomes)?

[jarmstropreamp]

The work with these molecules is early enough, that to my knowledge, there are no studies of this type that are published. And, in fact one could pose the same question for linear mRNA and ncRNA in exosomes (or ESVs). Most of the studies are on miRNA in ESVs, however we have done considerably work internally on characterizing long- mRNA and ncRNA in exosomes. These vesicles are fascinating!

[i000]

Circular RNAs are really "hot" these days. But, one should keep in mind that the amount of virtually all circular RNAs is lower in cancer (compared to non-cancerous adjacent normal tissue), thus they are not very promising as biomarkers. Also, their biological role is unclear. We did comprehensive literature review on this topic and found nothing truly convincing.

[davemessina]

That some circRNAs are lower in cancer than in normal tissue is a reliably detectable difference and is precisely what makes them an excellent biomarker.

There's additional evidence shown here: "Using circular RNA as a novel type of biomarker in the screening of gastric cancer". http://dx.doi.org/10.1016/j.cca.2015.02.018

[i000]

I am aware of this study. Funny, how they show expression in deltaCT to make it appear higher in cancer ;). But regardless, the marker has a specificity of 0.62 - really that excellent? Such a high false-positive rate does not belong anywhere near a patient sample.

[jarmstropreamp]

Our goal is to add additional markers to current candidates using our enrichment tech. It helps to remember that up until the ILMN/Solexa machine came online around 2005-6, we could not reliably detect (statistically) MAFs in heterogenous tumor samples below about 20% on the 454 platform (depending on how much cash you threw at the sample). Point is, our enrichment platform allows a deep dive into signals that have yet to be identified as reliable and low variance biomarkers.

[bhickey]

Are there some key publications you could point us at? I've been out of the bioimformatics game for a couple of years and I'm always curious to see where the state of the art is at.

[davemessina]

Here's a good place to start: http://dx.doi.org/10.1038/cr.2015.82

[danieltillett]

Dave what is the patent landscape like for circular RNA? I hope it is better than RNAi or antisense?

[davemessina]

Hi Daniel,

We've filed patents on our work.

Is there any biotech patent landscape as bad as RNAi? :)

[danieltillett]

Not that I know of :)

Good luck with your venture.