Beneath the indigo exterior of the Plex-ID is a level of sophistication we welcome at Omnica. The scope of the project was enormous, and it required over two years to perform the majority of the work. The result was a system that is more than an elaborate mechanism or pleasing enclosure design. It is a device that performs precise functions, and works as planned every time. The is the "desalter". It consists of a microplate stacker, carousel, and a fluidics system gantry.
DNA desalting in the Plex-ID is a patented process, but we can explain it and hopefully not oversimplify. Before the process begins, the unidentified DNA is amplified and combined with reactive primers. Afterwards it is delivered from a laboratory in sealed 96-well microplates in which each well contains the DNA and a distinctive reactive primer. A technician stages the microplates in the Plex-ID microplate stacker.
and Robert Fish (now retired) supplied electronics engineering,
and design support for the Plex-ID.
The number of microplates required for DNA identification varies, depending on how much the lab knows about the sample at the initial stage of the identification process. The ensuing desalting procedure will mix the amplified DNA with magnetic beads. Over a period of eleven minutes, DNA clinging to the beads is cleaned, isolated, and finally analyzed by the mass spectrometer.
The desalter mechanism consists of a gantry with sample needles and a carousel populated with 22 cuvettes, or mixing tubes. First, a gantry-mounted needle extracts magnetic beads from an onboard container. It then aspirates the DNA from a microplate well and the combination is deposited into one of the 22 cuvettes on the carousel. The needle is cleansed and moments later the carousel turns to position the next cuvette, where it waits for its infusion of magnetic beads and a dose of DNA.
This graphic of the desalter platform shows
some of the custom components we developed
for the project.
The DNA in the cuvette attaches to the magnetic beads, and during the following ten minute processing cycle the genetic material is thoroughly cleansed. Each cuvette has both an attached motor that spins the tiny collecting tube, and a set of outboard high-energy magnets, which are deployed at specified times during the process. It is at these moments, when the magnets are moved into position to attract and hold the beads in place along the inside wall of the cuvette, a gantry needle extracts the waste fluid. Rinse-spin-extraction cycles thoroughly cleanse the waste solutions from the cuvette, leaving only the beads with their attached DNA.
Finally, the DNA is separated from the beads, aspirated from the cuvette, and analyzed by the on-board time-of-flight mass spectrometer. The empty cuvette is cleaned again, and later rotates into position to be infused with fresh beads and a new DNA sample.
The mechanics and electronics were developed concurrently, but the logic and how the components were controlled by the system firmware was an assignment for our electronics department. This is where fine tuning of the system came together. Kevin Oberkramer orchestrated the hardware and firmware team as one of the leaders. He wanted to talk with us about his role in developing the desalter system, and offer insight regarding our activities beyond industrial design and mechanical engineering.
OMNIview: What were you asked to do for the Plex-ID project, and what design latitude did you have to accomplish your goals?
Kevin Oberkramer: We were tasked with two primary goals. One was to shrink it. The previous iteration of the device, the T-5000, was half the size of a room, and we needed to put that functionality into something much smaller. The carousel was a way we could multi-task some of the hardware and do parallel sample processing in minimal space. At the same time we were asked to substantially reduce the bill of materials cost by substituting custom parts for expensive off-the-shelf parts.
Another big challenge was meeting their throughput goal. The cleaning process is ten minutes, and they wanted to get one sample into the mass spectrometer every thirty seconds. The carousel, with multiple cuvettes, helped solve that problem.
As for design latitude, Ibis certainly gave us constraints, which changed as they refined the chemistry of the desalting process. They locked that down, but we controlled the mechanics of the procedure. Ibis told us what needed to be done, and then allowed us to manage how to do it. They were willing to let us brainstorm, design, and use our hardware philosophy to find answers. Essentially, they supported us in almost everything we did. Ibis had a budget and a technology to be implemented, but they let us use our creativity to find answers for the task at hand.
OMNIview: The desalter carousel is a complex mechanism. Was it difficult to develop this system, integrate all of this high-technology, and still be sure it would work as planned?
Kevin: Yes, to answer your question it was a lot of work, but it wasn't overly difficult, just time consuming. Along with the vacuum pumps and the solenoids, there is a long list of hardware. There are 33 microcontrollers, 54 motors, 67 PCBs, and many sensors. That's a lot of moving parts to keep track of. Not only that, everything has to be perfectly synchronized to get all the mechanisms to work together so the system runs like a clock and won't physically crash.
OMNIview: That is very interesting. What design methods did you use to make sure everything would operate smoothly and would not fail?
Kevin: We used a CAN-bus, a very robust multi-drop serial communications architecture that is similar to the interface used in automobiles. This is how all of our 33 mechanism controllers communicate – through these types of signal lines.
The recovery we built into the system addresses what happens if something goes wrong. In the same CAN-bus architecture, each independent controller knows its own job, watches for its own errors, and monitors failure conditions in all the other controllers. An advantage of the way we set this up is that it's a simple architecture that can run almost entirely independent from a PC.
Instead of using a single brain [a PC to run everything], the set of operations each controller is tasked to do is managed by a highly accurate clock inside every controller. All of these internal clocks are synchronized by a master clock controller, which sends out a forced global re-sync signal every second.
The error detection system is based on these timings. We require each controller to send a status message to the others every 100 milliseconds. They each know what the others are doing every tenth of a second, and if something goes wrong, a pause message is sent from every controller to all the others. The pause signal can result from something simple like running out of reagent fluid or because of a major event like a motor freezing.
In practice, a processor usually finds its own error in less than a tenth of a second and it sends the pause command. If the error is fixable, a controller can send a recovery command to get the system running again. But if a processor dies, all the controllers know very quickly that they did not receive a status message from the dead processor. In this case they all send a pause signal. That's the failsafe.
OMNIview: Finally the mystery has been answered. Now we know what you and your colleagues do in the electronics department.
Kevin: It's good to have a chance to discuss the technology we are able to put under the hood. We deal with system design in a microscopic way, so there is no visible way to show what we do. People can see a circuit board, and they may be impressed, but we can't show them the firmware. Now, at least our customers have some idea of the types of efforts the electronics department works on.
OMNIview: The primary development is complete, but I see the unit parked next to your desk. Are you still working on the project?
Kevin: We have a fully working version of the desalter system, so we are able to assist with some of the sustaining engineering and support. I had not worked on it for almost a year, but presently I am making some requested firmware changes and fixing minor bugs. Also, in the past they mostly placed the instrument in government labs like the CDC [Centers for Disease Control and Prevention]. They have asked me to work on some additional features that may be required at other facilities.
In 2009 the Plex-ID won two prestigious awards. The Wall Street Journal recognized the Plex-ID as the 2009 "Innovation of the Year" Gold award. The online magazine, The Scientist.com said the Plex-ID was one of "The ten most exciting tools to hit the life sciences in 2009".