The last few presentations were about state-changing tiles. Most self-assembly simply takes passive objects that are thermodynamically predisposed to bind each other. Under the right circumstances, these objects will fall into the most favorable position by any of a multitude of routes. The “programming” is in designing the lowest energy state. DNA origami falls into this category. Even algorithmic self assembly like the “counter” I referenced in the last post are just falling into thermodynamic wells that are generated in sequence as the tiles assemble. The tiles don’t “know” that they are bound or not.

The state change tiles start in one state, and upon binding a partner they change state. In the new state, they may become available for further binding or release other bound tiles. This system is far more “programmable” because the order of operations (as well as the final thermodynamic minimum) can be encoded into the tiles. The example demonstrated by Jennifer Padilla shows a set of tiles of type A and B that generate ABBB, four tiles linked together. Critically, the system does not make ABBBB – groups of five tiles. So B is not simply associating blindly with other B. It is “counting” internally. While all the B tiles are originally the same, they are in different states.

Essentially, you mix A + B⁰ and you get AB¹. B⁰ binds B­¹ to become AB¹B² and so on to form AB¹B²B³. The three B tiles are identical in composition but B³ is inert. So the chain is terminated. This strikes me as a far richer platform to build complex, evolving structures.

There were other theoretical presentations by Lila Kari, Alexandra Keenan, Matthew Patitz but I admit I didn’t follow them as well. Mark Arnold’s approach to DNA computation was interesting, but I was distracted thinking about assembly.