Recent trends in technology are challenging engineers to configure products at ever smaller scales. At the nano-scale, biological protein machines are commonly chosen as a power-source for a broad-range of nano-devices. This paper explores the challenges in designing these and similar systems, such as improving the emergent system performance that arises from the interactions of many stochastic components. We develop a domain-independent methodology, using multi-agent simulations as a means of modeling and predicting emergent system behavior across scales and structure-behavior-function representations for understanding and navigating the resulting design space. This methodology is validated with an application of synthetic myosin motor design at the nanoscale, with simulation results aligning well with the macroscopic performance of myosin-powered muscular contractions. The multi-agent simulation is implemented with myosins modeled as agents, allowing for the virtual design and experimentation of synthetic myosins with altered structures and mechanochemical behaviors. Four myosin populations are designed and simulated, with their emergent system performance determined by aggregating the contributions of each myosin agent over time. Although the multi-agent simulation successfully recreates the emergent behaviors of the myosins, it is difficult to draw conclusions about how each structural variation influences aggregate performance. SBF representations of the system are then developed to describe how the aggregate performance of the system is explainable in terms of myosin behaviors, which map directly to altered myosin structures. It is then demonstrated how an engineer may utilize these representations and experimental results to reason about, and configure a myosin system with optimal performance. The methodology is domain-independent, ensuring its extendibility to similar complex systems while aiding a designer in simplifying a complex physical phenomenon to a design space consisting of only a few critical parameters. The methodology is particularly suited for complex systems with many parts operating stochastically across scales, and should prove invaluable for engineers facing the challenges of biological nanoscale design, where designs with unique properties require novel approaches or useful configurations in nature await discovery.