This Software Takes Us One Step Closer to Biological Nanomachines

Scientists at MIT have developed a computer model that allows them to create three-dimensional DNA shapes of unprecedented complexity (like 20-faced icosahedrons). Their system represents a significant step forward in the field of nanoscale biological engineering.


Researchers led by MIT biological engineer Mark Bathe describe their design program in the latest issue of Nature Communications. The video up top gives a good summary of the team's achievements, but we reached out to Bathe for some clarification on the claim that the designs in the video constitute "the most complicated 3D structures ever made form DNA." His feedback:

The computational algorithm we've developed now enables the design and prediction of 3D structures of nearly arbitrary DNA assemblies. Our previous approach was limited to brick-like objects of parallel DNA molecules that was rather restrictive. The cages you see, for example, as well as the rings, could not be treated by the previous approach. It is important to note that the structures shown were previously designed manually by other researchers, but our computational algorithm now enables their in silico design and 3D structure prediction, which is central to their function (same as the 3D structure of a protein is essential to its function in binding other proteins, etc.)


One of the foremost aims of the burgeoning field of nanoscale engineering is to deepen our understanding of this relationship between structure and function. Imagine a program capable of translating a desired shape (for example: a geometric capsule that opens and closes for drug delivery) into a DNA sequence that assumes that shape when synthesized in the lab. Such a program would not only show you what your capsule looks like, helping you visualize it at a more relatable scale, but perhaps tell you about its structural properties. Imagine stress-testing a microscopic object the way you might stress-test the wing-design for a plane. Says Bathe:

The nanometer-scale is challenging because, unlike macroscopic objects such as cars, pencils, and chairs, we cannot easily 'see' it and manipulate it. Instead, we need to rely on very advanced imaging technologies such as Atomic Force Microscopy, Transmission Electron Microscopy, etc., and even those are highly limited because they only allow us to visualize larger-scale features, and often introduce artifacts into the sample. Thus, much like we want to be able to design an airplane in its entirety in the computer because airplanes are very difficult and expensive to build, we want to be able to design DNA nanostructures in the computer because they're both expensive to build, but also very difficult to see (unlike airplanes :)

Read the full scientific study in the latest issue of Nature Communications.

Share This Story

Get our newsletter



What the hell? I solved a large chunk of this problem, in software, under a fairly large symmetry group, as a first year undergrad, and Ned Seeman's lab at NYU has been fabbing stuff like this (and far more complex stuff, like electromechanical DNA computers) for even longer.

Clearly we submitted our paper to the wrong journals, because it got rejected for not being a particularly interesting result.