Abstract: DNA-functionalized nanoparticles, regarded as the programmable atom equivalent, enable the realization of hierarchically self-assembled superstructures. The self-assembled superstructures possessing unique mechanic, optical and electronic properties have prospective applications in the field of energy conservation, catalysis and medical diagnostics. With the development of nanotechnology, non-uniformly DNA-functionalized nanoparticles with DNA strands regioselectively distributed on the surfaces have been successfully synthesized. Recently, by utilizing the non-uniformly DNA-functionalized nanoparticles, researchers have created nanoparticle superstructures with complex architecture in the lab, such as discrete planet-satellite nanostructures, one-dimensional nanoparticle chains, and even three-dimensional networks. However, the mechanism and design rules of self-assembly of non-uniformly DNA-functionalized nanoparticles remain to be explored. Herein, the coarse-grained model of non-uniformly DNA-functionalized nanoparticles is constructed, and molecular dynamics is utilized to simulate the self-assembly process of DNA-programmable nanoparticles. It is demonstrated that the non-uniformly DNA-functionalized nanoparticles self-assemble into branched or even network-like superstructures through the hybridization of complementary DNA strands. The geometrical model of self-assembled superstructures is proposed to predict the relative position and distribution of nanoparticles inside the superstructures. Sparked by the molecular polymerization, we further explored the effect of stoichiometric ratio on the self-assembly of nanoparticles. The stoichiometric ratio of nanoparticles has remarkable effects on both the architecture of superstructures and the kinetics of DNA-programmable self-assembly of nanoparticles. As the stoichiometric ratio increased from 1.0 to 5.7, the self-assembled superstructures switch from the spanning networks to discrete branched clusters.