A microfluidic DNA processor dubbed a "lab-on-chip" has been developed by RIT researchers, capable of not only computation but also reading and writing data stored within DNA [h/t RIT.edu]. The prototype device supports artificial neural network computations on data stored within DNA, specifically microfluidic solutions of manipulated DNA molecules. The capabilities of this DNA CPU also extend to the expected mathematical and non-linear calculations you want to see from a CPU, and it should be capable of networking functionality with other devices.
Amlan Ganguly, head of the computer engineering department at Kate Gleason College of Engineering at the Rochester Institute of Technology, leads the research with the help of his department and researchers from the University of Minnesota. One of their goals in pushing DNA computation and storage is to find a more sustainable alternative to present-day big data technologies.
“We proposed to represent numbers through concentrations of solutions containing specifically manipulated DNA molecules and computing operations as manipulation of DNA molecules—operations like addition and multiplication and other non-linear functions necessary for network computations can be performed. That is the bridge from storage to computation and using DNA as a vehicle to do the computation,” Ganguly told the RIT.
While RIT's new DNA lab-on-chip is an impressive feat, it's important to note that it's only the latest step of many toward a future of viable DNA computing. Last year, we covered Chinese researchers and their successful attempts at making programmable DNA. This January, we also covered Biomemory, a startup looking to break into the data center with DNA storage while also selling a somewhat overpriced 1KB DNA storage card.
With all this talk of "programmable DNA" and "DNA storage," it may sound like we're wandering into the space of mad scientists and other such sci-fi tropes. However, some solid, grounded reasons exist for pursuing DNA computing and DNA storage. Besides being a much more environmentally friendly alternative, DNA storage shows promise as being far more dense in its capacity— as high as 3-to-6 orders of magnitude more than SSDs, according to the original paper.
DNA's potential as a storage medium or even computational resource boils down to its natural structure and traits. DNA naturally comprises four base ATGC molecules (adenine, thymine, guanine, and cytosine). This should allow for more efficient data storage than the 0/1 base numbers required with binary— especially combined with its microscopic scale. The storage system used here seems to still map binary over DNA but uses the added complexity to make it more easily rewritable.
Ganguly's team made this microfluidic DNA storage/computing device to further push the future of DNA storage and computing forward. The more robust DNA computing offered here is positioned for use in commercial applications (like data centers) and medical applications (like biomedical devices or forensics).
In the long term, Ganguly also focused on DNA computing's potential to reduce the environmental impact of mass data storage— a sensible goal, considering that data centers compose around 1.3% of global electrical demand. Of course, the biomedical applications seem most promising at this time due to the bio- and networking compatibility of the tech.
However, DNA computation and storage also come with their own unique issues— namely, very slow operation (orders of magnitude or literal hours slower) and, thus, unfeasibly high latency. While the economics of DNA storage could prove surprisingly appealing for long-term storage specifically, practical use of it on a large scale in its current form would require a front-end of modern high-end hardware to make it usable.
