Preprint · experimentally validated
DNA Polycube Inverse Design
Working backward from a shape to the blocks that build it
The problem
Forward design in DNA nanotechnology is straightforward to state and hard to scale: you can simulate whether a given set of building blocks self-assembles into a given shape, but searching for which blocks and rules produce an arbitrary target shape — while avoiding every possible misassembled alternative — is a much harder combinatorial problem.
What I helped build
This project treats cube-shaped DNA nanostructures as universal building blocks and solves the inverse design problem directly: given a target 3D shape, a multiscale simulation platform combined with constrained optimization identifies the minimal set of distinct polycubes and the interaction rules between them needed to assemble that shape — and only that shape.
The approach was validated on both finite structures and periodic lattices, including a fractal Menger cube, and extends to out-of-equilibrium reconfiguration: assembled structures can be rearranged after the fact by introducing invader cubes that displace the ones already in place.
Why it matters
Minimizing the block count isn't just an efficiency detail — every additional distinct block is another opportunity for the system to misfold, so finding the truly minimal rule set is what makes complex, reliable 3D self-assembly practical rather than theoretical.