DNA-coated colloids are designed systems. You attach specific DNA sequences to particle surfaces so that complementary strands on different particles bind selectively. Particle A binds particle B; neither binds particle C. The specificity is the point — it allows you to program which crystal structures self-assemble by controlling who sticks to whom.
Note (2026-11-13; read against the paper's full text): the essay says the structure can be switched by changing polymer brush length, salt concentration and particle core material. In the paper the crystals are a family of binary body-centered tetragonal structures characterised by a continuous parameter C between 0 (isostructural to cesium chloride) and 1 (isostructural to copper-gold), with intermediate values in between, not a switch between the two; the nonspecific interactions are tuned by the molecular weight of the polymer brush on the particle surface, the particle diameter, and the mixing of complementary grafted DNA strands. For salt, the paper reports that three NaCl concentrations from roughly 50 to 500 mM had a minimal effect on the equilibrium C value, and I found only polystyrene particles in the experiments, so the core material was not varied. The paper reports that simulations with pair potentials accounting for steric and van der Waals interactions reproduce the experiments quantitatively.
The paper shows that the crystal structure is not determined by the DNA. Binary mixtures of same-sized particles with the same DNA-encoded binding partners form different crystal structures — cesium-chloride or copper-gold types — depending on the steric repulsion and van der Waals attraction between the particles. These are the “nonspecific” interactions: they act regardless of DNA sequence, between all particles equally.
The DNA determines the bonding partners. The nonspecific interactions determine the geometry. The designed element (specific binding) answers who crystallizes with whom. The undesigned element (steric and van der Waals forces) answers how they arrange. By tuning the nonspecific interactions — changing polymer brush length, salt concentration, particle core material — the authors switch between crystal structures without touching the DNA.
The structural point: in a system designed for specificity, the outcome is controlled by the background. The foreground (DNA binding) is necessary — without it, there is no crystallization. But conditional on binding, the crystal geometry is set by forces that the designer typically ignores. The background is not noise; it is the architect that the foreground enables.