(Not Quite Live From) Synthetic Biology 2.0, Part V :: Fin

First off, here is the link for Synthetic Biology 3.0, next year's meeting in Switzerland.

This year's meeting was impressive on many counts.  As I have noted already (Part II), there was a distinct change in the flavor of the presentations.  The first day started out with a Nobel Laureate, followed up by a potential (probable?) future Laureate.  There was a significant amount of money in the room, from corporate representatives of synthesis companies to venture capitalist Vinod Khosla.  With respect to the technical presentations, the sheer diversity of systems and applications compared to two years ago was remarkable.  People are playing with more organisms and more parts (here's the meeting agenda).  The number of genes combined in several of the talks was itself remarkable.  Medical applications are clearly coming down the pike.

Yet I found something lacking.  As in 2004, there was no mention this year of a critical set of tools required in any engineering field.  It may not be sexy, but test and measurement gear is what allows rapid comparison of prediction and experimental outcome.  Without sophisticated test gear, you have no Pentium, no 777, no Honda Element, no SpaceShipOne.  At the moment, while each experiment presented at SB 2.0 may be technically beautiful and impressive, they are primarily one-offs.  There is no common signal, and there is no common way to compare experiments in different organisms.  This will eventually be addressed through some sort of standardization, such as is being attempted with Biobricks.  Yet I have always found the common signal in the Biobricks standard to be confusing.  I forget what it was called originally, but now the input and output relationships of the parts are defined in Polymerases Per Second, or POPS, the number of polymerases running into, or out of, a genetic element in a second.

As I write this, I finally realize why I don't like POPS.  As Drew Endy describes it, POPS is a way to allow abstraction from the level of genes and specific proteins up to devices with a common reference.  I understand this story, and it makes sense to me given the constraints of the biological parts we have to work with.  But here's the thing: measuring POPS is presently exceptionally hard.  You can test each part in a framework that allows the measurement of POPS, probably using a fluorescent protein as an output signal, which is only vaguely quantitative.  It is also not a direct measure of POPS, as there is at least one layer of function between the number of RNA polymerases running down DNA and the number of proteins that get translated from RNA.   But it gets worse; how to you troubleshoot the entire circuit?  Where do you stick the multimeter probes on the fly to see why your circuit isn't behaving as expected?  You don't.  Instead, you resort to microarrays to check RNA expression levels or you use protein assays.  Until there is a magic "POPSometer", there won't be any way to examine a circuit in real time.  Fluorescent proteins will never adequately fill this role, 1) because of the time required to fold and produce a fluorescent signal and 2) because you have to build a new circuit every time you want to stick the test probe in a new spot.

Moreover, tools presently in use provide the illusion to the uninitiated that the physical infrastructure of synthetic biology is already well developed.  It is fairly straightforward to get single cell fluorescence or behavior data at this point, but you have to presume the organism is running the program you wrote.  Separately, it is easy to sequence large amounts of DNA, generally purified from many individuals.  But you can't yet sequence a given bug behaving in a given way to make sure it is following the DNA you put into it.  And readily available sequencing technologies average over variation present in a population that may be critical to understanding function.

This technological mismatch extends to discussions of security.  We heard descriptions of various programs to monitor DNA synthesis efforts, which would tie into a surveillance network using a microbial background signal for the environment.  The later would serve as a reference for efforts to detect novel, and perhaps threatening organisms, in real time.  But there isn't yet any technology that can provide that sort of environmental information, nor will one be available in short order from what I have seen.

In summary, we are still at the beginning of a very long road.  Before chemical engineering came synthetic chemistry, and before biological engineering will come synthetic biology.  I just wish the community had better perspective on how far we have to go.