LMP Seminar: Inverse design of self-assembling systems: from lattices to capsids and multifarious structures

One of the major challenges of self-assembly is to design interactions between building blocks such that they assemble into a target shape and avoid competing alternative conformations. We have developed a multiscale molecular modeling approach, where we combine patchy particle simulations (abstracting each block as a sphere with patchy sites) and nucleotide-level coarse-grained modeling with oxDNA to design DNA nanostructures that self-assemble into multicomponent structures. The molecular modeling is coupled to our optimization algorithm, called SAT-assembly, which can quickly scan the design space of possible interactions and exclude those that lead to kinetic traps and alternative free-energy minima. We show applications of our method to realize highly-coveted pyrochlore lattice, polycube structures, and capsids using the smallest amount of building blocks required to avoid competing assemblies. We investigate the relation between assembly kinetics and the complexity of the solution in terms of the different number of building blocks and interaction sites. Next, we show the application of our framework to multifarious designs: systems which can form multiple desired target structures. We show its applications to design of a reconfigurable net, which can form different stored folded shapes. We further design multifarious nanostructures, where one set of DNA cubic-shaped building blocks can fold into multiple possible target shapes. Besides the theoretical framework, we present experimental proof-of-principle realizations of the designs that have been driven by the theoretical modeling: pyrochlore lattices assembled out of DNA wireframe origami, multifarious polycube nanostructures, and self-assembled polyhedras.